@piagent/platform 1.6.0 → 1.7.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (736) hide show
  1. package/CHANGELOG.md +104 -0
  2. package/README.md +24 -15
  3. package/README.vi.md +36 -2
  4. package/adapters/common/assurance-producers.json +109 -0
  5. package/adapters/common/structured-assurance-facts.mjs +153 -0
  6. package/adapters/node-typescript/contract-families.json +2811 -0
  7. package/adapters/node-typescript/contract-families.md +90 -0
  8. package/benchmarks/production-v1/grade.mjs +20 -0
  9. package/benchmarks/production-v1/spend-control.v1.json +27 -3
  10. package/benchmarks/production-v1/suite.json +5 -1
  11. package/benchmarks/production-v2/grade.mjs +892 -0
  12. package/benchmarks/production-v2/project/README.md +7 -0
  13. package/benchmarks/production-v2/project/config/service.json +4 -0
  14. package/benchmarks/production-v2/project/data/audit-ledger.jsonl +1 -0
  15. package/benchmarks/production-v2/project/docs/ops.md +3 -0
  16. package/benchmarks/production-v2/project/logs/incident.log +2 -0
  17. package/benchmarks/production-v2/project/package.json +11 -0
  18. package/benchmarks/production-v2/project/scripts/check.mjs +21 -0
  19. package/benchmarks/production-v2/project/src/backend/auth.js +3 -0
  20. package/benchmarks/production-v2/project/src/backend/billing-window.js +4 -0
  21. package/benchmarks/production-v2/project/src/backend/cache.js +15 -0
  22. package/benchmarks/production-v2/project/src/backend/invoice.js +4 -0
  23. package/benchmarks/production-v2/project/src/backend/revocation-cache.js +3 -0
  24. package/benchmarks/production-v2/project/src/data/csv.js +3 -0
  25. package/benchmarks/production-v2/project/src/data/dedup.js +5 -0
  26. package/benchmarks/production-v2/project/src/data/migration.js +9 -0
  27. package/benchmarks/production-v2/project/src/data/ndjson-stream.js +5 -0
  28. package/benchmarks/production-v2/project/src/data/versioned-replay.js +8 -0
  29. package/benchmarks/production-v2/project/src/frontend/chat-events.js +7 -0
  30. package/benchmarks/production-v2/project/src/frontend/pagination.js +8 -0
  31. package/benchmarks/production-v2/project/src/frontend/request-lifecycle.js +15 -0
  32. package/benchmarks/production-v2/project/src/frontend/search-state.js +8 -0
  33. package/benchmarks/production-v2/project/src/frontend/unicode-search.js +7 -0
  34. package/benchmarks/production-v2/project/src/fullstack/contract-sync.js +10 -0
  35. package/benchmarks/production-v2/project/src/platform/args.js +10 -0
  36. package/benchmarks/production-v2/project/src/platform/config.js +7 -0
  37. package/benchmarks/production-v2/project/src/platform/workflow-session.js +9 -0
  38. package/benchmarks/production-v2/project/src/platform/workspace.js +13 -0
  39. package/benchmarks/production-v2/project/src/reliability/checkpoint.js +5 -0
  40. package/benchmarks/production-v2/project/src/reliability/expiry.js +5 -0
  41. package/benchmarks/production-v2/project/src/reliability/retry.js +15 -0
  42. package/benchmarks/production-v2/project/test/smoke.test.js +21 -0
  43. package/benchmarks/production-v2/project/vendor/README.md +1 -0
  44. package/benchmarks/production-v2/prompts/abort-reconnect-supersession.md +11 -0
  45. package/benchmarks/production-v2/prompts/backend-frontend-contract-sync.md +12 -0
  46. package/benchmarks/production-v2/prompts/billing-cutoff-clock-skew.md +11 -0
  47. package/benchmarks/production-v2/prompts/bounded-retry.md +7 -0
  48. package/benchmarks/production-v2/prompts/chunked-record-boundary.md +9 -0
  49. package/benchmarks/production-v2/prompts/cli-double-dash.md +7 -0
  50. package/benchmarks/production-v2/prompts/config-precedence.md +6 -0
  51. package/benchmarks/production-v2/prompts/destructive-history-refusal.md +3 -0
  52. package/benchmarks/production-v2/prompts/expiry-boundary.md +7 -0
  53. package/benchmarks/production-v2/prompts/idempotent-replay-conflict.md +11 -0
  54. package/benchmarks/production-v2/prompts/incident-diagnosis.md +4 -0
  55. package/benchmarks/production-v2/prompts/invoice-rounding.md +9 -0
  56. package/benchmarks/production-v2/prompts/journeys/config-precedence-scout.md +3 -0
  57. package/benchmarks/production-v2/prompts/journeys/expiry-boundary-scout.md +3 -0
  58. package/benchmarks/production-v2/prompts/journeys/quoted-csv-scout.md +3 -0
  59. package/benchmarks/production-v2/prompts/journeys/recover-after-uncertain-send.md +4 -0
  60. package/benchmarks/production-v2/prompts/journeys/repository-prompt-injection-scout.md +4 -0
  61. package/benchmarks/production-v2/prompts/journeys/resumable-checkpoint-partial-failure-scout.md +4 -0
  62. package/benchmarks/production-v2/prompts/journeys/stable-dedup-scout.md +3 -0
  63. package/benchmarks/production-v2/prompts/journeys/stale-search-response-scout.md +3 -0
  64. package/benchmarks/production-v2/prompts/journeys/tenant-cache-isolation-scout.md +4 -0
  65. package/benchmarks/production-v2/prompts/journeys/verify-current-contract.md +4 -0
  66. package/benchmarks/production-v2/prompts/journeys/workflow-switch-same-session-scout.md +4 -0
  67. package/benchmarks/production-v2/prompts/pagination-boundary.md +7 -0
  68. package/benchmarks/production-v2/prompts/protected-env-refusal.md +2 -0
  69. package/benchmarks/production-v2/prompts/quoted-csv.md +6 -0
  70. package/benchmarks/production-v2/prompts/reconnect-chat-event-order.md +19 -0
  71. package/benchmarks/production-v2/prompts/repository-prompt-injection.md +5 -0
  72. package/benchmarks/production-v2/prompts/resumable-checkpoint-partial-failure.md +11 -0
  73. package/benchmarks/production-v2/prompts/revoked-session-cache.md +10 -0
  74. package/benchmarks/production-v2/prompts/schema-migration.md +6 -0
  75. package/benchmarks/production-v2/prompts/stable-dedup.md +6 -0
  76. package/benchmarks/production-v2/prompts/stale-search-response.md +6 -0
  77. package/benchmarks/production-v2/prompts/tenant-cache-isolation.md +6 -0
  78. package/benchmarks/production-v2/prompts/tenant-role-authorization.md +6 -0
  79. package/benchmarks/production-v2/prompts/unicode-search.md +6 -0
  80. package/benchmarks/production-v2/prompts/workflow-switch-same-session.md +30 -0
  81. package/benchmarks/production-v2/prompts/workspace-order.md +7 -0
  82. package/benchmarks/production-v2/spend-control.v1.json +115 -0
  83. package/benchmarks/production-v2/suite.json +1009 -0
  84. package/benchmarks/production-v2/variant.mjs +169 -0
  85. package/benchmarks/production-v3/CONTRACT-REVISION-20260922.md +31 -0
  86. package/benchmarks/production-v3/calibration/corpus.v1.json +63 -0
  87. package/benchmarks/production-v3/calibration/human-review-rubric.v1.json +36 -0
  88. package/benchmarks/production-v3/grade.mjs +1017 -0
  89. package/benchmarks/production-v3/project/README.md +7 -0
  90. package/benchmarks/production-v3/project/config/service.json +4 -0
  91. package/benchmarks/production-v3/project/data/audit-ledger.jsonl +1 -0
  92. package/benchmarks/production-v3/project/docs/ops.md +3 -0
  93. package/benchmarks/production-v3/project/logs/incident.log +2 -0
  94. package/benchmarks/production-v3/project/package.json +11 -0
  95. package/benchmarks/production-v3/project/scripts/check.mjs +21 -0
  96. package/benchmarks/production-v3/project/src/backend/auth.js +3 -0
  97. package/benchmarks/production-v3/project/src/backend/billing-window.js +4 -0
  98. package/benchmarks/production-v3/project/src/backend/cache.js +15 -0
  99. package/benchmarks/production-v3/project/src/backend/invoice.js +4 -0
  100. package/benchmarks/production-v3/project/src/backend/revocation-cache.js +12 -0
  101. package/benchmarks/production-v3/project/src/data/csv.js +3 -0
  102. package/benchmarks/production-v3/project/src/data/dedup.js +5 -0
  103. package/benchmarks/production-v3/project/src/data/migration.js +9 -0
  104. package/benchmarks/production-v3/project/src/data/ndjson-stream.js +5 -0
  105. package/benchmarks/production-v3/project/src/data/versioned-replay.js +8 -0
  106. package/benchmarks/production-v3/project/src/frontend/chat-events.js +7 -0
  107. package/benchmarks/production-v3/project/src/frontend/pagination.js +8 -0
  108. package/benchmarks/production-v3/project/src/frontend/request-lifecycle.js +23 -0
  109. package/benchmarks/production-v3/project/src/frontend/search-state.js +8 -0
  110. package/benchmarks/production-v3/project/src/frontend/unicode-search.js +7 -0
  111. package/benchmarks/production-v3/project/src/fullstack/contract-sync.js +10 -0
  112. package/benchmarks/production-v3/project/src/platform/args.js +10 -0
  113. package/benchmarks/production-v3/project/src/platform/config.js +7 -0
  114. package/benchmarks/production-v3/project/src/platform/workflow-session.js +9 -0
  115. package/benchmarks/production-v3/project/src/platform/workspace.js +13 -0
  116. package/benchmarks/production-v3/project/src/reliability/checkpoint.js +5 -0
  117. package/benchmarks/production-v3/project/src/reliability/expiry.js +14 -0
  118. package/benchmarks/production-v3/project/src/reliability/retry.js +15 -0
  119. package/benchmarks/production-v3/project/test/smoke.test.js +21 -0
  120. package/benchmarks/production-v3/project/vendor/README.md +1 -0
  121. package/benchmarks/production-v3/prompts/abort-reconnect-supersession.md +11 -0
  122. package/benchmarks/production-v3/prompts/backend-frontend-contract-sync.md +20 -0
  123. package/benchmarks/production-v3/prompts/billing-cutoff-clock-skew.md +11 -0
  124. package/benchmarks/production-v3/prompts/bounded-retry.md +7 -0
  125. package/benchmarks/production-v3/prompts/chunked-record-boundary.md +11 -0
  126. package/benchmarks/production-v3/prompts/cli-double-dash.md +7 -0
  127. package/benchmarks/production-v3/prompts/config-precedence.md +6 -0
  128. package/benchmarks/production-v3/prompts/destructive-history-refusal.md +9 -0
  129. package/benchmarks/production-v3/prompts/expiry-boundary.md +14 -0
  130. package/benchmarks/production-v3/prompts/idempotent-replay-conflict.md +13 -0
  131. package/benchmarks/production-v3/prompts/incident-diagnosis.md +4 -0
  132. package/benchmarks/production-v3/prompts/invoice-rounding.md +9 -0
  133. package/benchmarks/production-v3/prompts/journeys/config-precedence-scout.md +3 -0
  134. package/benchmarks/production-v3/prompts/journeys/expiry-boundary-scout.md +3 -0
  135. package/benchmarks/production-v3/prompts/journeys/quoted-csv-scout.md +3 -0
  136. package/benchmarks/production-v3/prompts/journeys/recover-after-uncertain-send.md +4 -0
  137. package/benchmarks/production-v3/prompts/journeys/repository-prompt-injection-scout.md +4 -0
  138. package/benchmarks/production-v3/prompts/journeys/resumable-checkpoint-partial-failure-scout.md +4 -0
  139. package/benchmarks/production-v3/prompts/journeys/stable-dedup-scout.md +3 -0
  140. package/benchmarks/production-v3/prompts/journeys/stale-search-response-scout.md +3 -0
  141. package/benchmarks/production-v3/prompts/journeys/tenant-cache-isolation-scout.md +4 -0
  142. package/benchmarks/production-v3/prompts/journeys/verify-current-contract.md +4 -0
  143. package/benchmarks/production-v3/prompts/journeys/workflow-switch-same-session-scout.md +4 -0
  144. package/benchmarks/production-v3/prompts/pagination-boundary.md +9 -0
  145. package/benchmarks/production-v3/prompts/protected-env-refusal.md +8 -0
  146. package/benchmarks/production-v3/prompts/quoted-csv.md +6 -0
  147. package/benchmarks/production-v3/prompts/reconnect-chat-event-order.md +20 -0
  148. package/benchmarks/production-v3/prompts/repository-prompt-injection.md +5 -0
  149. package/benchmarks/production-v3/prompts/resumable-checkpoint-partial-failure.md +15 -0
  150. package/benchmarks/production-v3/prompts/revoked-session-cache.md +10 -0
  151. package/benchmarks/production-v3/prompts/schema-migration.md +6 -0
  152. package/benchmarks/production-v3/prompts/stable-dedup.md +6 -0
  153. package/benchmarks/production-v3/prompts/stale-search-response.md +6 -0
  154. package/benchmarks/production-v3/prompts/tenant-cache-isolation.md +6 -0
  155. package/benchmarks/production-v3/prompts/tenant-role-authorization.md +6 -0
  156. package/benchmarks/production-v3/prompts/unicode-search.md +6 -0
  157. package/benchmarks/production-v3/prompts/workflow-switch-same-session.md +30 -0
  158. package/benchmarks/production-v3/prompts/workspace-order.md +7 -0
  159. package/benchmarks/production-v3/public-checks/abort-reconnect-supersession.js +34 -0
  160. package/benchmarks/production-v3/public-checks/expiry-boundary.js +29 -0
  161. package/benchmarks/production-v3/public-checks/revoked-session-cache.js +20 -0
  162. package/benchmarks/production-v3/spend-control.v1.json +121 -0
  163. package/benchmarks/production-v3/suite.json +1010 -0
  164. package/benchmarks/production-v3/variant.mjs +178 -0
  165. package/catalog/capabilities.json +16 -16
  166. package/docs/authority-manifest.md +1 -1
  167. package/docs/auto-delegation-policy.md +38 -73
  168. package/docs/benchmark-codex-mcp-metadata-contract-v1.md +159 -0
  169. package/docs/benchmark-codex-mcp-metadata-contract-v2.md +182 -0
  170. package/docs/benchmark-codex-mcp-metadata-contract-v3.md +194 -0
  171. package/docs/benchmark-public-contract-addendum-v2.md +85 -0
  172. package/docs/benchmark-public-contract-addendum-v3.md +116 -0
  173. package/docs/command-reference-vietnamese.md +30 -35
  174. package/docs/context-engine.md +5 -3
  175. package/docs/context-window-policy.md +37 -1
  176. package/docs/distribution-standard.md +9 -9
  177. package/docs/en/architecture.md +26 -13
  178. package/docs/languages.json +4 -4
  179. package/docs/operator-manual-vietnamese.md +26 -46
  180. package/docs/piagent-codex-benchmark-recovery-plan.md +1817 -0
  181. package/docs/piagent-core-evaluation-specs-2026-09-06.json +706 -0
  182. package/docs/piagent-core-improvement-plan.md +528 -0
  183. package/docs/piagent-core-offline-implementation-2026-09-06.md +219 -0
  184. package/docs/piagent-core-plan-checkpoint-2026-09-06.json +69 -0
  185. package/docs/piagent-core-reasoning-research-2026-09-06.md +225 -0
  186. package/docs/piagent-core-research-sources-2026-09-06.json +416 -0
  187. package/docs/piagent-open-session-wait.md +19 -0
  188. package/docs/piagent-runtime-benchmark-execution-plan.md +685 -0
  189. package/docs/publishing-for-teams.md +7 -0
  190. package/docs/quality-benchmark.md +299 -23
  191. package/docs/quickstart-vietnamese.md +2 -2
  192. package/docs/release-install-policy.md +32 -9
  193. package/docs/research/harness-next-evidence-review.md +113 -0
  194. package/docs/runtime-harness-standard.md +3 -3
  195. package/docs/security-threat-model.md +1 -1
  196. package/docs/subagent-orchestration-capabilities.md +19 -51
  197. package/docs/subagents-and-multiagent.md +21 -47
  198. package/docs/task-implementation-contract.md +3 -3
  199. package/docs/task-lifecycle-vietnamese.md +1 -1
  200. package/docs/team-onboarding.md +7 -7
  201. package/docs/third-party-neutrality.md +7 -0
  202. package/docs/usage-observability.md +1 -1
  203. package/docs/vi/architecture.md +26 -13
  204. package/docs-site/architecture.html +1 -1
  205. package/docs-site/benchmark.html +89 -95
  206. package/docs-site/commands.html +1 -1
  207. package/docs-site/content/benchmark.html +88 -94
  208. package/docs-site/content/en/benchmark.html +134 -90
  209. package/docs-site/content/en/index.html +19 -19
  210. package/docs-site/content/en/runtime.html +5 -5
  211. package/docs-site/content/en/whats-new.html +23 -1
  212. package/docs-site/content/en/workflows.html +2 -2
  213. package/docs-site/content/index.html +19 -19
  214. package/docs-site/content/runtime.html +6 -6
  215. package/docs-site/content/whats-new.html +23 -1
  216. package/docs-site/content/workflows.html +2 -2
  217. package/docs-site/context-engine.html +1 -1
  218. package/docs-site/en/architecture.html +1 -1
  219. package/docs-site/en/benchmark.html +135 -91
  220. package/docs-site/en/commands.html +1 -1
  221. package/docs-site/en/context-engine.html +1 -1
  222. package/docs-site/en/index.html +20 -20
  223. package/docs-site/en/mcp-approval.html +1 -1
  224. package/docs-site/en/mcp-auth.html +1 -1
  225. package/docs-site/en/mcp-servers.html +1 -1
  226. package/docs-site/en/mcp.html +1 -1
  227. package/docs-site/en/permissions.html +1 -1
  228. package/docs-site/en/profiles.html +1 -1
  229. package/docs-site/en/quickstart.html +1 -1
  230. package/docs-site/en/runtime.html +6 -6
  231. package/docs-site/en/scope.html +1 -1
  232. package/docs-site/en/security.html +1 -1
  233. package/docs-site/en/team.html +1 -1
  234. package/docs-site/en/whats-new.html +24 -2
  235. package/docs-site/en/workflows.html +3 -3
  236. package/docs-site/index.html +20 -20
  237. package/docs-site/mcp-approval.html +1 -1
  238. package/docs-site/mcp-auth.html +1 -1
  239. package/docs-site/mcp-servers.html +1 -1
  240. package/docs-site/mcp.html +1 -1
  241. package/docs-site/permissions.html +1 -1
  242. package/docs-site/profiles.html +1 -1
  243. package/docs-site/quickstart.html +1 -1
  244. package/docs-site/runtime.html +7 -7
  245. package/docs-site/scope.html +1 -1
  246. package/docs-site/security.html +1 -1
  247. package/docs-site/team.html +1 -1
  248. package/docs-site/whats-new.html +24 -2
  249. package/docs-site/workflows.html +3 -3
  250. package/evals/architecture-conformance-v1/lane.json +18 -0
  251. package/evals/architecture-conformance-v1/runner.mjs +77 -0
  252. package/evals/fixtures/approved-host-contracts.invalid.json +43 -0
  253. package/evals/fixtures/approved-host-contracts.valid.json +43 -0
  254. package/evals/fixtures/benchmark-assurance-evidence.invalid.json +1 -1
  255. package/evals/fixtures/benchmark-assurance-evidence.valid.json +1 -1
  256. package/evals/fixtures/benchmark-attempt-outcome.invalid.json +42 -0
  257. package/evals/fixtures/benchmark-attempt-outcome.valid.json +42 -0
  258. package/evals/fixtures/benchmark-independent-verification-plan.invalid.json +7 -0
  259. package/evals/fixtures/benchmark-independent-verification-plan.valid.json +21 -0
  260. package/evals/fixtures/codex-relative-efficiency-v1.invalid.json +4 -0
  261. package/evals/fixtures/codex-relative-efficiency-v1.valid.json +94 -0
  262. package/evals/fixtures/composite-criterion.invalid.json +10 -0
  263. package/evals/fixtures/composite-criterion.valid.json +33 -0
  264. package/evals/fixtures/contract-family-library.invalid.json +42 -0
  265. package/evals/fixtures/contract-family-library.valid.json +41 -0
  266. package/evals/fixtures/contract-selection-recipe.invalid.json +29 -0
  267. package/evals/fixtures/contract-selection-recipe.valid.json +29 -0
  268. package/evals/fixtures/helper-request.invalid.json +1 -1
  269. package/evals/fixtures/helper-request.valid.json +1 -1
  270. package/evals/fixtures/host-contract-plan.invalid.json +42 -0
  271. package/evals/fixtures/host-contract-plan.valid.json +41 -0
  272. package/evals/fixtures/piagent-webui/subagent-tree-v1.valid.json +7 -0
  273. package/evals/fixtures/scoped-project-verification-receipt-v2.invalid.json +42 -0
  274. package/evals/fixtures/scoped-project-verification-receipt-v2.valid.json +42 -0
  275. package/evals/fixtures/scoped-project-verification-receipt-v3.invalid.json +57 -0
  276. package/evals/fixtures/scoped-project-verification-receipt-v3.valid.json +57 -0
  277. package/evals/fixtures/scoped-verification-receipt-v1.invalid.json +40 -0
  278. package/evals/fixtures/scoped-verification-receipt-v1.valid.json +40 -0
  279. package/evals/fixtures/task-authority-snapshot.valid.json +4 -4
  280. package/evals/fs4-readiness-matrix.v1.json +2 -2
  281. package/evals/fs5-artifact-archive/README.md +32 -0
  282. package/evals/fs5-artifact-archive/manifest.json +148 -0
  283. package/evals/fs5-artifact-archive/sha256/1f073f6d42c093a3182685e0e096144c8b3befd87594aeea1af05ce17653860c.txt +380 -0
  284. package/evals/fs5-artifact-archive/sha256/1ff46984c5ff668647dfad94248cd9dcfb72384a8aa529279345582f90d7486a.txt +331 -0
  285. package/evals/fs5-artifact-archive/sha256/20a5e1ef1cab3234a888b876b3cec52e702a05bc5ac000b691f94c651beed9ca.txt +800 -0
  286. package/evals/fs5-artifact-archive/sha256/29bb79e96b047d79056626d747baa0b3a6ca9d4da7ad80b9668c0085e5581706.txt +116 -0
  287. package/evals/fs5-artifact-archive/sha256/2bcc0cf6c18eee0c080c30e727610b0f10a278f5862855d920ba5797f07a2f2d.txt +252 -0
  288. package/evals/fs5-artifact-archive/sha256/37ecc65febe2c5947ac9288ad958bc149e0f08c6beda137edb1c660908853832.txt +324 -0
  289. package/evals/fs5-artifact-archive/sha256/5d937632d2cfff34dbbb781dd367bb78feb64c36d571c36ca15f2cc40c576af8.txt +142 -0
  290. package/evals/fs5-artifact-archive/sha256/6057c7bf2f40f8e3db7e293f08d745133c05293b75e9f363fb4e2e53bd09a76a.txt +244 -0
  291. package/evals/fs5-artifact-archive/sha256/62ea354afb6816b148ae9bd95ffb1ee5e29580eb4a22473e1ae19dbb471ec36b.txt +930 -0
  292. package/evals/fs5-artifact-archive/sha256/8987775b742f3b3650d75cce84e6d58fcdd3dfb835d03dac4c2bfe6d9eea3a2e.txt +903 -0
  293. package/evals/fs5-artifact-archive/sha256/902e886d0920b66811f54d0d1d933d70f8fab8704813b46d578bee6af1f21eef.txt +942 -0
  294. package/evals/fs5-artifact-archive/sha256/976a9908adfd765ee7be0655616f39a44fff64824f5e8535d726082e550b02eb.txt +910 -0
  295. package/evals/fs5-artifact-archive/sha256/9c07bda5bfe7bb7b0959844ccea6790735cc9f154d8d5972e18785148cea6d9a.txt +101 -0
  296. package/evals/fs5-artifact-archive/sha256/b936a604db87a697bf69581c33d753e26d9e5e2ccce40bf9c9b2070948f49919.txt +931 -0
  297. package/evals/fs5-artifact-archive/sha256/bf91f69473f48fc3de7c7c55e7ce893ba0511074c4ec930d9f3bcb9615d8e9d5.txt +950 -0
  298. package/evals/fs5-artifact-archive/sha256/ce4ed1b4a033f2a8cb4632d1336b0867edb3454ce1380f0642c4055149e6ac59.txt +784 -0
  299. package/evals/fs5-artifact-archive/sha256/db97f444db8ba604512daf0aa26b3fe1e2a120ace2b21d4730db5b18aa0c528f.txt +950 -0
  300. package/evals/fs5-causal-arm.v1.json +10 -6
  301. package/evals/fs5-pilot-protocol.v1.json +3 -3
  302. package/evals/fs5-pilot-protocol.v2.json +8 -8
  303. package/evals/fs5-pilot-protocol.v3.json +8 -8
  304. package/evals/fs5-pilot-protocol.v4.json +7 -7
  305. package/evals/fs5-pilot-protocol.v5.json +10 -8
  306. package/evals/harness-next/development-corpus.mjs +106 -0
  307. package/evals/harness-next/resource-budget-probe.mjs +38 -0
  308. package/evals/harness-next/run-calibration.mjs +83 -0
  309. package/evals/harness-next/run-resource-probe.mjs +55 -0
  310. package/evals/harness-next/selected-development-corpus.mjs +27 -0
  311. package/evals/harness-next.v1.json +51 -0
  312. package/evals/long-horizon-v1/README.md +19 -0
  313. package/evals/long-horizon-v1/runner.mjs +116 -48
  314. package/evals/long-horizon-v1/worker.mjs +77 -30
  315. package/evals/private-holdout-v1/CURRENT_BOUNDARY.md +53 -0
  316. package/evals/private-holdout-v1/README.md +5 -2
  317. package/evals/private-holdout-v1/public-exposure.v1.json +1 -1
  318. package/evals/private-holdout-v1/public-exposure.v2.json +273 -0
  319. package/evals/runtime-conformance-v1/README.md +43 -0
  320. package/evals/runtime-conformance-v1/evaluate-paid-canary.mjs +198 -0
  321. package/evals/runtime-conformance-v1/lane.json +57 -0
  322. package/evals/runtime-conformance-v1/paid-canary.json +31 -0
  323. package/evals/runtime-conformance-v1/runner.mjs +456 -0
  324. package/package.json +9 -4
  325. package/packages/piagent-core/README.md +15 -6
  326. package/packages/piagent-core/benchmark/benchmark-all-attempt-efficiency.js +55 -0
  327. package/packages/piagent-core/benchmark/benchmark-artifact-binding.js +49 -0
  328. package/packages/piagent-core/benchmark/benchmark-bootstrap.js +265 -14
  329. package/packages/piagent-core/benchmark/benchmark-budget-governor.js +271 -0
  330. package/packages/piagent-core/benchmark/benchmark-campaign.js +592 -0
  331. package/packages/piagent-core/benchmark/benchmark-candidate.js +8 -3
  332. package/packages/piagent-core/benchmark/benchmark-checkpoint.js +67 -0
  333. package/packages/piagent-core/benchmark/benchmark-claim-restrictions.js +269 -2
  334. package/packages/piagent-core/benchmark/benchmark-cli.js +120 -3
  335. package/packages/piagent-core/benchmark/benchmark-codex-events.js +235 -0
  336. package/packages/piagent-core/benchmark/benchmark-codex-outcome.js +20 -0
  337. package/packages/piagent-core/benchmark/benchmark-codex-relative-efficiency.js +699 -0
  338. package/packages/piagent-core/benchmark/benchmark-codex-rollout.js +521 -0
  339. package/packages/piagent-core/benchmark/benchmark-codex.js +104 -8
  340. package/packages/piagent-core/benchmark/benchmark-comparison.js +284 -1
  341. package/packages/piagent-core/benchmark/benchmark-core.js +111 -110
  342. package/packages/piagent-core/benchmark/benchmark-diagnostic-source-evidence.js +40 -0
  343. package/packages/piagent-core/benchmark/benchmark-diagnostic-treatment.js +104 -0
  344. package/packages/piagent-core/benchmark/benchmark-evaluator-v3.js +226 -0
  345. package/packages/piagent-core/benchmark/benchmark-execution-guard.js +23 -1
  346. package/packages/piagent-core/benchmark/benchmark-forensics.js +67 -52
  347. package/packages/piagent-core/benchmark/benchmark-goal-assessment.js +193 -0
  348. package/packages/piagent-core/benchmark/benchmark-goal-report.js +87 -0
  349. package/packages/piagent-core/benchmark/benchmark-independent-verification-observation.js +64 -0
  350. package/packages/piagent-core/benchmark/benchmark-matrix-summary.js +141 -0
  351. package/packages/piagent-core/benchmark/benchmark-matrix.js +1 -0
  352. package/packages/piagent-core/benchmark/benchmark-normalized-cost.js +62 -0
  353. package/packages/piagent-core/benchmark/benchmark-pi-home.js +1 -0
  354. package/packages/piagent-core/benchmark/benchmark-preflight.js +70 -7
  355. package/packages/piagent-core/benchmark/benchmark-process.js +23 -9
  356. package/packages/piagent-core/benchmark/benchmark-production-completion.js +84 -0
  357. package/packages/piagent-core/benchmark/benchmark-production-stage-validation.js +209 -0
  358. package/packages/piagent-core/benchmark/benchmark-provider-free-evidence.js +462 -0
  359. package/packages/piagent-core/benchmark/benchmark-provider-free-owner.js +24 -0
  360. package/packages/piagent-core/benchmark/benchmark-provider-wire.js +378 -0
  361. package/packages/piagent-core/benchmark/benchmark-record-validation.d.ts +2 -0
  362. package/packages/piagent-core/benchmark/benchmark-record-validation.js +343 -8
  363. package/packages/piagent-core/benchmark/benchmark-release-claim-controls.js +132 -0
  364. package/packages/piagent-core/benchmark/benchmark-report.js +94 -5
  365. package/packages/piagent-core/benchmark/benchmark-resume-recovery.js +280 -10
  366. package/packages/piagent-core/benchmark/benchmark-runner-policy.js +88 -1
  367. package/packages/piagent-core/benchmark/benchmark-runtime-causal.js +244 -0
  368. package/packages/piagent-core/benchmark/benchmark-runtime-identity.js +50 -1
  369. package/packages/piagent-core/benchmark/benchmark-runtime.js +102 -10
  370. package/packages/piagent-core/benchmark/benchmark-safety-evidence.js +74 -0
  371. package/packages/piagent-core/benchmark/benchmark-service-tier.js +189 -0
  372. package/packages/piagent-core/benchmark/benchmark-stage-diagnostic.js +246 -33
  373. package/packages/piagent-core/benchmark/benchmark-suite-identity.js +264 -1
  374. package/packages/piagent-core/benchmark/benchmark-suite-runtime.js +4 -0
  375. package/packages/piagent-core/benchmark/benchmark-suite.js +232 -4
  376. package/packages/piagent-core/benchmark/benchmark-summary-support.js +33 -1
  377. package/packages/piagent-core/benchmark/benchmark-timing-contract.js +189 -0
  378. package/packages/piagent-core/benchmark/benchmark-timing-diagnostics.js +110 -139
  379. package/packages/piagent-core/benchmark/benchmark-transport-evidence.js +203 -0
  380. package/packages/piagent-core/benchmark/benchmark-transport-observation.js +147 -0
  381. package/packages/piagent-core/benchmark/benchmark-tree-identity.js +2 -1
  382. package/packages/piagent-core/benchmark/benchmark-usage.js +410 -37
  383. package/packages/piagent-core/benchmark/benchmark-workflow.js +136 -38
  384. package/packages/piagent-core/benchmark/fs5-pilot-protocol.js +21 -2
  385. package/packages/piagent-core/capabilities/runtime-integrity.js +6 -0
  386. package/packages/piagent-core/extensions/acceptance-api-baseline.js +24 -0
  387. package/packages/piagent-core/extensions/acceptance-assessment.js +153 -0
  388. package/packages/piagent-core/extensions/acceptance-authenticated-admission.d.ts +41 -0
  389. package/packages/piagent-core/extensions/acceptance-authenticated-admission.js +120 -0
  390. package/packages/piagent-core/extensions/acceptance-behavior-proof.js +1000 -0
  391. package/packages/piagent-core/extensions/acceptance-boundary-binding.js +31 -0
  392. package/packages/piagent-core/extensions/acceptance-boundary-guidance.js +214 -0
  393. package/packages/piagent-core/extensions/acceptance-callable-scanner.js +244 -0
  394. package/packages/piagent-core/extensions/acceptance-clause-group.js +45 -0
  395. package/packages/piagent-core/extensions/acceptance-composite-contract.js +334 -0
  396. package/packages/piagent-core/extensions/acceptance-contract-conjunction.js +78 -0
  397. package/packages/piagent-core/extensions/acceptance-contract-selection.js +161 -0
  398. package/packages/piagent-core/extensions/acceptance-contract-semantics.js +211 -211
  399. package/packages/piagent-core/extensions/acceptance-declared-api.js +126 -0
  400. package/packages/piagent-core/extensions/acceptance-diagnostic-policy.ts +14 -0
  401. package/packages/piagent-core/extensions/acceptance-document-content.js +73 -0
  402. package/packages/piagent-core/extensions/acceptance-durable-execution.d.ts +59 -0
  403. package/packages/piagent-core/extensions/acceptance-durable-execution.js +139 -0
  404. package/packages/piagent-core/extensions/acceptance-error-classes.js +1 -1
  405. package/packages/piagent-core/extensions/acceptance-evidence-store.js +223 -0
  406. package/packages/piagent-core/extensions/acceptance-executable-evidence.js +490 -88
  407. package/packages/piagent-core/extensions/acceptance-execution-diagnostics.js +34 -0
  408. package/packages/piagent-core/extensions/acceptance-execution-snapshot.js +174 -0
  409. package/packages/piagent-core/extensions/acceptance-executor/.dockerignore +21 -0
  410. package/packages/piagent-core/extensions/acceptance-executor/Dockerfile +5 -0
  411. package/packages/piagent-core/extensions/acceptance-executor/README.md +103 -0
  412. package/packages/piagent-core/extensions/acceptance-executor/budget.mjs +54 -0
  413. package/packages/piagent-core/extensions/acceptance-executor/callback-intrinsics.mjs +126 -0
  414. package/packages/piagent-core/extensions/acceptance-executor/capabilities.mjs +127 -0
  415. package/packages/piagent-core/extensions/acceptance-executor/guest.mjs +285 -0
  416. package/packages/piagent-core/extensions/acceptance-executor/intrinsics.mjs +133 -0
  417. package/packages/piagent-core/extensions/acceptance-executor/invocation.mjs +39 -0
  418. package/packages/piagent-core/extensions/acceptance-executor/jobs.mjs +29 -0
  419. package/packages/piagent-core/extensions/acceptance-executor/module-graph.mjs +70 -0
  420. package/packages/piagent-core/extensions/acceptance-executor/node-profile.mjs +398 -0
  421. package/packages/piagent-core/extensions/acceptance-executor/package-lock.json +40 -0
  422. package/packages/piagent-core/extensions/acceptance-executor/package.json +10 -0
  423. package/packages/piagent-core/extensions/acceptance-executor/protocol.mjs +248 -0
  424. package/packages/piagent-core/extensions/acceptance-executor/reference-identity.mjs +61 -0
  425. package/packages/piagent-core/extensions/acceptance-executor/references.mjs +14 -0
  426. package/packages/piagent-core/extensions/acceptance-executor/structured-clone.mjs +53 -0
  427. package/packages/piagent-core/extensions/acceptance-executor/values.mjs +97 -0
  428. package/packages/piagent-core/extensions/acceptance-executor/worker.mjs +75 -0
  429. package/packages/piagent-core/extensions/acceptance-expiry-calendar-proof.js +162 -0
  430. package/packages/piagent-core/extensions/acceptance-expression-parser.js +155 -0
  431. package/packages/piagent-core/extensions/acceptance-focused-verification.js +40 -0
  432. package/packages/piagent-core/extensions/acceptance-host-configuration.js +337 -0
  433. package/packages/piagent-core/extensions/acceptance-identifier-criteria.js +25 -0
  434. package/packages/piagent-core/extensions/acceptance-independent-contract.js +209 -0
  435. package/packages/piagent-core/extensions/acceptance-independent-registry.js +106 -0
  436. package/packages/piagent-core/extensions/acceptance-input-provenance.js +40 -0
  437. package/packages/piagent-core/extensions/acceptance-integer-domains.js +322 -0
  438. package/packages/piagent-core/extensions/acceptance-invalid-input-receipt.js +20 -0
  439. package/packages/piagent-core/extensions/acceptance-invalid-partitions.js +26 -0
  440. package/packages/piagent-core/extensions/acceptance-invalid-sentinel-proof.js +477 -0
  441. package/packages/piagent-core/extensions/acceptance-isolated-executor.js +254 -0
  442. package/packages/piagent-core/extensions/acceptance-lexical-sentinels.js +130 -0
  443. package/packages/piagent-core/extensions/acceptance-literal-dataflow.js +735 -0
  444. package/packages/piagent-core/extensions/acceptance-local-binding.js +106 -0
  445. package/packages/piagent-core/extensions/acceptance-normalizer-evidence.js +719 -0
  446. package/packages/piagent-core/extensions/acceptance-obligations.js +75 -0
  447. package/packages/piagent-core/extensions/acceptance-parameter-contract.js +756 -0
  448. package/packages/piagent-core/extensions/acceptance-precedence-contract.js +572 -0
  449. package/packages/piagent-core/extensions/acceptance-proof-capability.js +26 -0
  450. package/packages/piagent-core/extensions/acceptance-receipt.js +208 -180
  451. package/packages/piagent-core/extensions/acceptance-result-contract-evidence.js +159 -0
  452. package/packages/piagent-core/extensions/acceptance-source-origin.js +214 -0
  453. package/packages/piagent-core/extensions/acceptance-state-machine-evidence.js +64 -8
  454. package/packages/piagent-core/extensions/acceptance-tagged-event-evidence.js +685 -0
  455. package/packages/piagent-core/extensions/acceptance-temporal-contract.js +756 -0
  456. package/packages/piagent-core/extensions/acceptance-temporal-dataflow-proof.js +618 -0
  457. package/packages/piagent-core/extensions/acceptance-test-binding-evidence.js +20 -2
  458. package/packages/piagent-core/extensions/acceptance-test-bindings.js +86 -0
  459. package/packages/piagent-core/extensions/acceptance-tuple-shapes.js +12 -0
  460. package/packages/piagent-core/extensions/context-engine.js +12 -17
  461. package/packages/piagent-core/extensions/context-index-policy.js +4 -0
  462. package/packages/piagent-core/extensions/context-query-paths.js +57 -0
  463. package/packages/piagent-core/extensions/criterion-context-pack.js +109 -4
  464. package/packages/piagent-core/extensions/criterion-graph.js +4 -1
  465. package/packages/piagent-core/extensions/failure-types.ts +2 -2
  466. package/packages/piagent-core/extensions/fallback-base-policy.ts +101 -0
  467. package/packages/piagent-core/extensions/guard-types.ts +9 -1
  468. package/packages/piagent-core/extensions/piagent-guard.ts +374 -364
  469. package/packages/piagent-core/extensions/readonly-inline-inspection.ts +378 -0
  470. package/packages/piagent-core/extensions/recovery-decision-validation.ts +92 -0
  471. package/packages/piagent-core/extensions/shell-write-targets.js +15 -14
  472. package/packages/piagent-core/extensions/task-contract-view.js +4 -4
  473. package/packages/piagent-core/extensions/task-digest-migration.js +2 -2
  474. package/packages/piagent-core/extensions/task-journal.js +48 -3
  475. package/packages/piagent-core/extensions/task-lifecycle.js +32 -0
  476. package/packages/piagent-core/extensions/task-runtime-audit.js +8 -2
  477. package/packages/piagent-core/extensions/task-state.js +140 -43
  478. package/packages/piagent-core/extensions/task-terminal-disposition.js +10 -0
  479. package/packages/piagent-core/extensions/trajectory-contract.ts +5 -0
  480. package/packages/piagent-core/extensions/verification-intelligence.js +109 -20
  481. package/packages/piagent-core/extensions/workflow-catalog.ts +98 -0
  482. package/packages/piagent-core/extensions/workspace-evidence-roots.js +52 -16
  483. package/packages/piagent-core/extensions/workspace-revision.js +48 -0
  484. package/packages/piagent-core/extensions/workspace-verification-snapshot.js +13 -0
  485. package/packages/piagent-core/package.json +3 -3
  486. package/packages/piagent-core/policies/base-policy.json +43 -2
  487. package/packages/piagent-core/policy/authority-manifest.v1.json +3 -3
  488. package/packages/piagent-core/prompts/be-to-fe.md +2 -8
  489. package/packages/piagent-core/prompts/plan.md +1 -1
  490. package/packages/piagent-core/prompts/platform-improve.md +2 -6
  491. package/packages/piagent-core/prompts/review.md +2 -7
  492. package/packages/piagent-core/prompts/scout.md +1 -1
  493. package/packages/piagent-core/prompts/task.md +4 -3
  494. package/packages/piagent-core/runtime/context/adaptive-planner.ts +7 -3
  495. package/packages/piagent-core/runtime/hooks/agent-start-hook.ts +60 -59
  496. package/packages/piagent-core/runtime/hooks/agent-start-types.ts +48 -0
  497. package/packages/piagent-core/runtime/hooks/completion-hook.ts +83 -89
  498. package/packages/piagent-core/runtime/hooks/input-hook.ts +98 -7
  499. package/packages/piagent-core/runtime/hooks/protected-refusal-completion.ts +86 -0
  500. package/packages/piagent-core/runtime/hooks/session-hooks.ts +5 -2
  501. package/packages/piagent-core/runtime/hooks/session-start-hook.ts +150 -19
  502. package/packages/piagent-core/runtime/hooks/tool-result-hook.ts +111 -92
  503. package/packages/piagent-core/runtime/hooks/tool-result-metadata.ts +9 -0
  504. package/packages/piagent-core/runtime/hooks/tool-result-navigation.ts +36 -0
  505. package/packages/piagent-core/runtime/hooks/tool-result-retrieval.ts +42 -0
  506. package/packages/piagent-core/runtime/hooks/tool-result-value-helpers.ts +66 -0
  507. package/packages/piagent-core/runtime/input/chat-images.ts +1 -1
  508. package/packages/piagent-core/runtime/inspection/activity-event-adapter.ts +1 -1
  509. package/packages/piagent-core/runtime/inspection/bounded-emission-guard.ts +286 -0
  510. package/packages/piagent-core/runtime/inspection/runtime-event-store.ts +37 -7
  511. package/packages/piagent-core/runtime/inspection/same-process-bridge-proof.ts +1 -1
  512. package/packages/piagent-core/runtime/inspection/task-handoff-history.ts +2 -2
  513. package/packages/piagent-core/runtime/inspection/task-subagent-tree.ts +27 -4
  514. package/packages/piagent-core/runtime/inspection/tool-failure-classification.ts +69 -11
  515. package/packages/piagent-core/runtime/inspection/webui-snapshot.ts +104 -22
  516. package/packages/piagent-core/runtime/inspection/workspace-file-reader.ts +2 -132
  517. package/packages/piagent-core/runtime/model/provider-wire-fingerprint.ts +20 -0
  518. package/packages/piagent-core/runtime/model/service-tier-runtime.ts +242 -0
  519. package/packages/piagent-core/runtime/orchestration/helper-lifecycle.ts +16 -12
  520. package/packages/piagent-core/runtime/orchestration/minimal-delegation-policy.ts +71 -0
  521. package/packages/piagent-core/runtime/orchestration/owned-work-budget.ts +28 -20
  522. package/packages/piagent-core/runtime/orchestration/role-policy.ts +27 -9
  523. package/packages/piagent-core/runtime/orchestration/subagent-tool-policy.ts +329 -0
  524. package/packages/piagent-core/runtime/policy/block-remedy.ts +0 -4
  525. package/packages/piagent-core/runtime/policy/source-worktree-transaction.ts +75 -14
  526. package/packages/piagent-core/runtime/product/activity-inspector.ts +6 -3
  527. package/packages/piagent-core/runtime/product/edit-verifier-timing.ts +26 -0
  528. package/packages/piagent-core/runtime/product/efficiency-metrics.ts +15 -7
  529. package/packages/piagent-core/runtime/product/operator-projections.ts +8 -3
  530. package/packages/piagent-core/runtime/product/task-usage-hooks.ts +78 -0
  531. package/packages/piagent-core/runtime/product/task-usage.ts +104 -0
  532. package/packages/piagent-core/runtime/quality/edit-freshness-guard.ts +102 -0
  533. package/packages/piagent-core/runtime/quality/exact-output-contract.ts +20 -10
  534. package/packages/piagent-core/runtime/quality/performance-assurance.ts +3 -3
  535. package/packages/piagent-core/runtime/quality/protected-refusal-handoff.ts +35 -0
  536. package/packages/piagent-core/runtime/recovery/acceptance-recovery-guidance.ts +21 -0
  537. package/packages/piagent-core/runtime/recovery/continuation-budget.ts +12 -3
  538. package/packages/piagent-core/runtime/recovery/diagnostic-delivery.ts +61 -0
  539. package/packages/piagent-core/runtime/recovery/handoff-projection.ts +196 -58
  540. package/packages/piagent-core/runtime/recovery/handoff-verifier-binding.ts +24 -0
  541. package/packages/piagent-core/runtime/recovery/independent-verification-recovery.ts +79 -0
  542. package/packages/piagent-core/runtime/recovery/pristine-task-policy-refresh.ts +91 -0
  543. package/packages/piagent-core/runtime/recovery/recovery-policy.ts +31 -40
  544. package/packages/piagent-core/runtime/recovery/resume-state.ts +120 -19
  545. package/packages/piagent-core/runtime/recovery/semantic-repair-handshake.ts +7 -13
  546. package/packages/piagent-core/runtime/recovery/task-evidence-coverage-refresh.ts +217 -0
  547. package/packages/piagent-core/runtime/registration/activity-inspector-command.ts +2 -1
  548. package/packages/piagent-core/runtime/registration/context-commands.ts +28 -10
  549. package/packages/piagent-core/runtime/registration/knowledge-tools.ts +9 -5
  550. package/packages/piagent-core/runtime/registration/operator-catalogs.ts +4 -9
  551. package/packages/piagent-core/runtime/registration/operator-request-intake.ts +49 -0
  552. package/packages/piagent-core/runtime/registration/policy-tools.ts +8 -6
  553. package/packages/piagent-core/runtime/registration/session-commands.ts +10 -8
  554. package/packages/piagent-core/runtime/registration/session-wait-tool.ts +54 -0
  555. package/packages/piagent-core/runtime/registration/task-completion-tools.ts +39 -20
  556. package/packages/piagent-core/runtime/registration/task-evidence-tools.ts +21 -12
  557. package/packages/piagent-core/runtime/registration/task-start-conditional-mutation.ts +45 -0
  558. package/packages/piagent-core/runtime/registration/task-start-guidance.ts +24 -15
  559. package/packages/piagent-core/runtime/registration/task-start-manifest.ts +54 -0
  560. package/packages/piagent-core/runtime/registration/task-start-tool.ts +79 -79
  561. package/packages/piagent-core/runtime/registration/workflow-commands.ts +64 -93
  562. package/packages/piagent-core/runtime/runtime-limits.ts +16 -2
  563. package/packages/piagent-core/runtime/session/adaptive-context-analysis.ts +461 -0
  564. package/packages/piagent-core/runtime/session/adaptive-context-governor.ts +464 -0
  565. package/packages/piagent-core/runtime/session/adaptive-context-ledger.ts +265 -0
  566. package/packages/piagent-core/runtime/session/completion-signals.ts +23 -0
  567. package/packages/piagent-core/runtime/session/message-signals.ts +6 -18
  568. package/packages/piagent-core/runtime/session/operator-request-carry.ts +44 -0
  569. package/packages/piagent-core/runtime/session/retrieval-evidence-runtime.ts +188 -0
  570. package/packages/piagent-core/runtime/session/runtime-state.ts +37 -39
  571. package/packages/piagent-core/runtime/session/source-checkout-read-grants.ts +35 -0
  572. package/packages/piagent-core/runtime/session/system-prompt.ts +76 -14
  573. package/packages/piagent-core/runtime/session/task-contract-presentation.ts +39 -0
  574. package/packages/piagent-core/runtime/session/uncertain-send-continuation.ts +231 -0
  575. package/packages/piagent-core/runtime/session/usage.ts +4 -3
  576. package/packages/piagent-core/runtime/tools/apply-patch-tool.ts +6 -6
  577. package/packages/piagent-core/runtime/tools/phase-tool-runtime.ts +1 -1
  578. package/packages/piagent-core/runtime/tools/phase-tools.ts +10 -9
  579. package/packages/piagent-core/runtime/tools/session-wait.ts +42 -0
  580. package/packages/piagent-core/runtime/tools/tool-groups.ts +22 -5
  581. package/packages/piagent-core/runtime/trajectory/trajectory-runtime.ts +192 -5
  582. package/packages/piagent-core/runtime/trajectory/trajectory-types.ts +3 -2
  583. package/packages/piagent-core/runtime/verification/acceptance-api-baseline.js +488 -0
  584. package/packages/piagent-core/runtime/verification/completion-preparation.ts +22 -0
  585. package/packages/piagent-core/runtime/verification/composite-completion-finalizer.ts +105 -0
  586. package/packages/piagent-core/runtime/verification/composite-journal-anchor.ts +181 -0
  587. package/packages/piagent-core/runtime/verification/composite-runtime-binding.ts +76 -0
  588. package/packages/piagent-core/runtime/verification/composite-scoped-mediation.ts +318 -0
  589. package/packages/piagent-core/runtime/verification/composite-session-persistence.ts +175 -0
  590. package/packages/piagent-core/runtime/verification/composite-task-publication.ts +273 -0
  591. package/packages/piagent-core/runtime/verification/exact-verifier-reuse.ts +101 -0
  592. package/packages/piagent-core/runtime/verification/host-project-verification.ts +71 -0
  593. package/packages/piagent-core/runtime/verification/independent-acceptance-runtime.ts +500 -0
  594. package/packages/piagent-core/runtime/verification/runtime-assurance-facts.ts +500 -0
  595. package/packages/piagent-core/runtime/verification/runtime-contract-runner.ts +107 -0
  596. package/packages/piagent-core/runtime/verification/shell-verification-snapshots.ts +56 -0
  597. package/packages/piagent-core/runtime/workflows/change-clarification.ts +61 -0
  598. package/packages/piagent-core/runtime/workflows/input-routing.ts +21 -14
  599. package/packages/piagent-core/runtime/workflows/runtime-intake-compaction.ts +201 -0
  600. package/packages/piagent-core/runtime/workflows/task-followup-policy.ts +163 -0
  601. package/packages/piagent-core/runtime/workflows/task-intake.ts +276 -169
  602. package/packages/piagent-core/runtime/workflows/task-repair-intent.ts +67 -0
  603. package/packages/piagent-core/runtime/workflows/task-scope-resolution.ts +83 -0
  604. package/packages/piagent-core/runtime/workflows/webui-runtime-command.ts +5 -2
  605. package/packages/piagent-core/runtime/workflows/webui-workflow.ts +39 -32
  606. package/packages/piagent-core/runtime/workflows/workflow-follow-up.ts +33 -0
  607. package/packages/piagent-core/security/workspace-file-reader.ts +132 -0
  608. package/packages/piagent-core/subagents/piagent-oracle.md +5 -3
  609. package/packages/piagent-core/subagents/piagent-planner.md +6 -4
  610. package/packages/piagent-core/subagents/piagent-reviewer.md +5 -3
  611. package/packages/piagent-core/subagents/piagent-scout.md +4 -2
  612. package/packages/piagent-core/subagents/piagent-worker.md +6 -4
  613. package/packages/piagent-webui/dist/client/assets/{MarkdownMessage-CxeZqTte.js → MarkdownMessage-DGk34zJT.js} +1 -1
  614. package/packages/piagent-webui/dist/client/assets/{index-BV-_03yl.css → index-DVahUVWy.css} +1 -1
  615. package/packages/piagent-webui/dist/client/assets/index-DiKRmCEP.js +176 -0
  616. package/packages/piagent-webui/dist/client/index.html +2 -2
  617. package/packages/piagent-webui/extension/dispatch-user-entry.ts +13 -0
  618. package/packages/piagent-webui/extension/piagent-webui.ts +13 -3
  619. package/packages/piagent-webui/extension/review-controller.ts +29 -6
  620. package/packages/piagent-webui/extension/same-session-bridge.ts +7 -8
  621. package/packages/piagent-webui/extension/session-stream-adapter.ts +88 -16
  622. package/packages/piagent-webui/extension/source-mutation-controller.ts +29 -3
  623. package/packages/piagent-webui/extension/source-open-controller.ts +29 -7
  624. package/packages/piagent-webui/extension/source-revert-controller.ts +26 -3
  625. package/packages/piagent-webui/gateway/gateway-events.ts +22 -5
  626. package/packages/piagent-webui/gateway/gateway-service.ts +7 -2
  627. package/packages/piagent-webui/gateway/gateway-session-stream.ts +170 -32
  628. package/packages/piagent-webui/gateway/rpc-ui-context.ts +34 -9
  629. package/packages/piagent-webui/gateway/runtime-command-controller.ts +46 -6
  630. package/packages/piagent-webui/gateway/runtime-session-controls.ts +17 -2
  631. package/packages/piagent-webui/gateway/session-catalog.ts +9 -3
  632. package/packages/piagent-webui/gateway/session-command-controller.ts +29 -9
  633. package/packages/piagent-webui/gateway/session-effective-options.ts +2 -9
  634. package/packages/piagent-webui/gateway/session-inspection-registry.ts +11 -2
  635. package/packages/piagent-webui/gateway/session-operation-lifecycle.ts +237 -0
  636. package/packages/piagent-webui/gateway/session-operation-retry-control.ts +38 -0
  637. package/packages/piagent-webui/gateway/session-operation-settlement.ts +28 -0
  638. package/packages/piagent-webui/gateway/session-operation-watchdog.ts +9 -2
  639. package/packages/piagent-webui/gateway/session-prompt-dispatch.ts +37 -0
  640. package/packages/piagent-webui/gateway/session-runtime-close.ts +36 -0
  641. package/packages/piagent-webui/gateway/session-runtime-factory.ts +183 -6
  642. package/packages/piagent-webui/gateway/session-runtime-ownership.ts +48 -0
  643. package/packages/piagent-webui/gateway/session-runtime-state.ts +24 -0
  644. package/packages/piagent-webui/gateway/session-runtime-supervisor.ts +118 -117
  645. package/packages/piagent-webui/ownership/session-lease-store.ts +5 -2
  646. package/packages/piagent-webui/server/core-inspection-provider.ts +75 -15
  647. package/packages/piagent-webui/server/loopback-server.ts +9 -7
  648. package/packages/piagent-webui/server/terminal-delivery-receipt.ts +144 -0
  649. package/packages/piagent-webui/server/transcript-projection.ts +150 -11
  650. package/packages/piagent-webui/shared/message-correlation.ts +25 -0
  651. package/schemas/approved-host-contracts.schema.json +973 -0
  652. package/schemas/benchmark-attempt-outcome.schema.json +303 -0
  653. package/schemas/benchmark-independent-verification-plan.schema.json +34 -0
  654. package/schemas/codex-relative-efficiency-v1.schema.json +155 -0
  655. package/schemas/composite-criterion.schema.json +788 -0
  656. package/schemas/contract-family-library.schema.json +28 -0
  657. package/schemas/contract-selection-recipe.schema.json +31 -0
  658. package/schemas/failure-evidence.schema.json +2 -2
  659. package/schemas/helper-request.schema.json +16 -4
  660. package/schemas/host-contract-plan.schema.json +25 -0
  661. package/schemas/piagent-webui/gateway-protocol-v1.schema.json +16 -2
  662. package/schemas/piagent-webui/runtime-command-v1.schema.json +26 -3
  663. package/schemas/piagent-webui/runtime-event-v2.schema.json +46 -5
  664. package/schemas/piagent-webui/session-live-state-v1.schema.json +2 -0
  665. package/schemas/piagent-webui/snapshot-v1.schema.json +3 -0
  666. package/schemas/piagent-webui/source-change-v1.schema.json +12 -0
  667. package/schemas/piagent-webui/subagent-tree-v1.schema.json +11 -1
  668. package/schemas/piagent-webui/transcript-v1.schema.json +2 -1
  669. package/schemas/project-profile.schema.json +1 -1
  670. package/schemas/public-contracts/codex-mcp-metadata-v1.schema.json +206 -0
  671. package/schemas/public-contracts/codex-mcp-metadata-v2.schema.json +228 -0
  672. package/schemas/public-contracts/codex-mcp-metadata-v3.schema.json +233 -0
  673. package/schemas/public-contracts/revocation-v1.schema.json +87 -0
  674. package/schemas/public-contracts/subscription-contract-v1.schema.json +76 -0
  675. package/schemas/scoped-project-verification-receipt-v2.schema.json +83 -0
  676. package/schemas/scoped-project-verification-receipt-v3.schema.json +295 -0
  677. package/schemas/scoped-verification-receipt-v1.schema.json +76 -0
  678. package/schemas/task-contract.schema.json +45 -0
  679. package/scripts/approve-independent-verification.mjs +57 -0
  680. package/scripts/benchmark-arm-observer.mjs +320 -0
  681. package/scripts/benchmark-budget-ipc.mjs +34 -0
  682. package/scripts/benchmark-budget-launcher.mjs +91 -0
  683. package/scripts/benchmark-budget-processes.mjs +133 -0
  684. package/scripts/benchmark-budget-publication.mjs +228 -0
  685. package/scripts/benchmark-budget-runtime.mjs +136 -0
  686. package/scripts/benchmark-candidate-ablation.mjs +265 -0
  687. package/scripts/benchmark-codex-journey.mjs +406 -0
  688. package/scripts/benchmark-independent-verification.mjs +628 -0
  689. package/scripts/benchmark-journey-outcome.mjs +92 -0
  690. package/scripts/benchmark-production-v3-human-review.mjs +192 -0
  691. package/scripts/benchmark-public-input-catalog.mjs +158 -0
  692. package/scripts/benchmark-registered-plan-author.mjs +355 -0
  693. package/scripts/benchmark-registered-plan-recipe-support.mjs +85 -0
  694. package/scripts/benchmark-registered-plan-recipes-backend.mjs +165 -0
  695. package/scripts/benchmark-registered-plan-recipes-data.mjs +164 -0
  696. package/scripts/benchmark-registered-plan-recipes-platform.mjs +165 -0
  697. package/scripts/benchmark-runner-completion.mjs +133 -0
  698. package/scripts/benchmark-runner-configuration.mjs +159 -0
  699. package/scripts/benchmark-runner-core.mjs +224 -387
  700. package/scripts/benchmark-runner-finalization.mjs +144 -27
  701. package/scripts/benchmark-runner-provider-boundary.mjs +44 -0
  702. package/scripts/benchmark-runner-schedule.mjs +256 -0
  703. package/scripts/benchmark-runner-support.mjs +136 -4
  704. package/scripts/benchmark-runner.mjs +95 -18
  705. package/scripts/benchmark-scoped-broker-custody.mjs +483 -0
  706. package/scripts/benchmark-scoped-docs-verifier.mjs +203 -0
  707. package/scripts/benchmark-scoped-frozen-qualification.mjs +676 -0
  708. package/scripts/benchmark-scoped-pi-router.mjs +196 -0
  709. package/scripts/benchmark-scoped-project-verification-receipt.mjs +49 -0
  710. package/scripts/benchmark-scoped-project-verifier.mjs +461 -0
  711. package/scripts/benchmark-scoped-session-custody.mjs +599 -0
  712. package/scripts/benchmark-scoped-session-factory.mjs +342 -0
  713. package/scripts/benchmark-scoped-tool-broker.mjs +523 -0
  714. package/scripts/benchmark-scoped-verification-supervisor.mjs +773 -0
  715. package/scripts/benchmark-session-evaluator.mjs +169 -0
  716. package/scripts/benchmark-session-provider-boundary.mjs +62 -0
  717. package/scripts/benchmark-session-usage.mjs +28 -0
  718. package/scripts/benchmark-session.mjs +343 -50
  719. package/scripts/benchmark-webui-journey.mjs +678 -0
  720. package/scripts/check-public-wording.mjs +46 -0
  721. package/scripts/configure-subagents.sh +42 -34
  722. package/scripts/fs4-readiness-evaluation.mjs +7 -1
  723. package/scripts/helper-evaluation.mjs +1 -1
  724. package/scripts/phase-tools-evaluation.mjs +1 -0
  725. package/scripts/pi-auto.sh +7 -1
  726. package/scripts/pi-usage-history.mjs +201 -11
  727. package/scripts/piagent-cli.mjs +8 -0
  728. package/scripts/private-holdout-readiness.mjs +7 -1
  729. package/scripts/public-evaluation-exposure.mjs +140 -0
  730. package/scripts/select-independent-verification.mjs +62 -0
  731. package/scripts/team-doctor.sh +1 -1
  732. package/scripts/typescript-loader.mjs +135 -1
  733. package/scripts/update-global.mjs +6 -7
  734. package/scripts/verify-local.sh +41 -55
  735. package/scripts/verify-release-identity.mjs +5 -3
  736. package/packages/piagent-webui/dist/client/assets/index-DynIMe8I.js +0 -176
@@ -0,0 +1,500 @@
1
+ import { createHash } from "node:crypto";
2
+ import fs from "node:fs";
3
+ import path from "node:path";
4
+ import { fileURLToPath } from "node:url";
5
+ import type { ExtensionContext } from "@earendil-works/pi-coding-agent";
6
+ import type { TaskContract } from "../../extensions/guard-types.ts";
7
+ import type { AuthenticatedAssessment } from "../../extensions/acceptance-authenticated-admission.js";
8
+ import { openHostContractConfiguration, independentRequestAdmissionBlock } from "../../extensions/acceptance-host-configuration.js";
9
+ import { captureCompositeExecutionSnapshot } from "../../extensions/acceptance-execution-snapshot.js";
10
+ import { registerIndependentAcceptanceProvider } from "../../extensions/acceptance-independent-registry.js";
11
+ import { durableTaskContractMatches, workingTreeSnapshot } from "../../extensions/task-state.js";
12
+ import { workingTreeEvidenceDigest } from "../../extensions/task-lifecycle.js";
13
+ import { captureWorkspaceVerificationSnapshot } from "../../extensions/workspace-verification-snapshot.js";
14
+ import { openCompositeAssuranceJournal, openCompositeAssuranceRuntime, sealCompositePlanContext } from "./runtime-assurance-facts.ts";
15
+ import { compareConfigDocumentLiterals, compareIncidentClaims, comparePolicyRefusalOutput } from "../../../../adapters/common/structured-assurance-facts.mjs";
16
+ import { createCompositeCodeChildCollector, createRuntimeContractRunner } from "./runtime-contract-runner.ts";
17
+ import { capturePersistedAssistantResponse, commitCompositeWebUiDelivery, persistedAssistantResponseObservation, webUiDeliveryObservation } from "./composite-session-persistence.ts";
18
+ import { openScopedMediationEvidence, scopedMediationFactObservation } from "./composite-scoped-mediation.ts";
19
+ import { compositeTaskPublicationDigest, openCompositeTaskPublicationStore, recoverCompositeTaskPublication } from "./composite-task-publication.ts";
20
+ import { assertCompositeCriterion, compositePhaseHeadDigest, compositeSettlementTaskStatus, compositeTaskContractDigest, projectCompositeLifecycle } from "./composite-runtime-binding.ts";
21
+ import type { RuntimeSessionState } from "../session/runtime-state.ts";
22
+ import type { CompletionPreparation } from "./completion-preparation.ts";
23
+ type ResponseObservation = { origin: "assistant"; bytes: string }; type DeferredCompletionFinalizer = { preflight: () => object | false; publication: (capability: object) => object | false; finalize: (capability: object) => boolean };
24
+ type Options = { state: RuntimeSessionState; installedRoot: string; configPath?: string;
25
+ activeTask: (ctx: ExtensionContext) => TaskContract | undefined; authorizeSourceRead: (ctx: ExtensionContext, sourcePath: string) => boolean; writeTask?: (cwd: string, task: TaskContract) => TaskContract };
26
+ type CompositeItem = { contract: any; plan: any; declarations: any; producer: any; authority: any;
27
+ receipts: Map<string, any>; codeChildren: Map<string, ReturnType<typeof createCompositeCodeChildCollector>>; preparation?: object;
28
+ aggregatePreparation?: any; aggregate?: object; publication?: object };
29
+ type Entry = { configuration?: ReturnType<typeof openHostContractConfiguration>; dispose: () => void;
30
+ controller: AbortController; block?: string; stop?: { reason: string; attemptId?: string }; hasComposite: boolean;
31
+ compositeReason?: string; composites: CompositeItem[];
32
+ response?: ResponseObservation; persistedResponse?: object; persistenceError?: string;
33
+ finalizer?: DeferredCompletionFinalizer; provisional?: boolean;
34
+ runners: Array<{ contract: any; runner: ReturnType<typeof createRuntimeContractRunner>; receipt?: AuthenticatedAssessment }>;
35
+ pending?: Promise<void> };
36
+ const STRUCTURED_MODULE = fileURLToPath(new URL("../../../../adapters/common/structured-assurance-facts.mjs", import.meta.url)); const ID = /^[A-Za-z0-9][A-Za-z0-9:._~-]{0,159}$/;
37
+ const sha = (value: string | Buffer) => createHash("sha256").update(value).digest("hex");
38
+ const exact = (value: any, fields: string[]) => value && typeof value === "object" && !Array.isArray(value)
39
+ && Object.keys(value).length === fields.length && fields.every(field => Object.hasOwn(value, field));
40
+ function stableFile(file: string, maximum: number): Buffer {
41
+ if (fs.realpathSync.native(file) !== file) throw new Error("Producer path is not canonical");
42
+ const descriptor = fs.openSync(file, fs.constants.O_RDONLY | fs.constants.O_NOFOLLOW);
43
+ try {
44
+ const before = fs.fstatSync(descriptor);
45
+ if (!before.isFile() || before.nlink !== 1 || before.size < 1 || before.size > maximum) throw new Error("Producer file is unsafe");
46
+ const bytes = fs.readFileSync(descriptor), after = fs.fstatSync(descriptor), atPath = fs.lstatSync(file);
47
+ if (bytes.length !== before.size || atPath.isSymbolicLink()
48
+ || ["dev", "ino", "mode", "nlink", "size", "mtimeMs", "ctimeMs"].some(field => before[field] !== after[field] || after[field] !== atPath[field])) {
49
+ throw new Error("Producer file changed during read");
50
+ }
51
+ return bytes;
52
+ } finally { fs.closeSync(descriptor); }
53
+ }
54
+ function loadStructuredProducer(installedRoot: string) {
55
+ const root = fs.realpathSync.native(installedRoot), manifestPath = path.join(root, "adapters/common/assurance-producers.json");
56
+ const modulePath = path.join(root, "adapters/common/structured-assurance-facts.mjs");
57
+ if (fs.realpathSync.native(modulePath) !== fs.realpathSync.native(STRUCTURED_MODULE)) throw new Error("Loaded producer is not the installed producer");
58
+ const manifestBytes = stableFile(manifestPath, 1024 * 1024), moduleBytes = stableFile(modulePath, 512 * 1024);
59
+ const manifest = JSON.parse(manifestBytes.toString("utf8"));
60
+ if (!exact(manifest, ["schemaVersion", "manifestVersion", "suiteContract", "status", "authority", "activation",
61
+ "producerResultsMayGrantPass", "producers", "qualification"])
62
+ || manifest.schemaVersion !== 1 || manifest.manifestVersion !== "assurance-producers-v1"
63
+ || manifest.suiteContract !== "production-v2-da2" || manifest.status !== "provider-free-qualified"
64
+ || manifest.authority !== "host-runtime-content-facts" || manifest.activation !== "host-composite-content-v1"
65
+ || manifest.producerResultsMayGrantPass !== false || !Array.isArray(manifest.producers) || manifest.producers.length !== 1
66
+ || !exact(manifest.qualification, ["state", "approvedDesignVectorCount", "runtimeActivation", "remainingConjunctsAreHostAuthenticated"])
67
+ || manifest.qualification.state !== "provider-free-qualified" || manifest.qualification.approvedDesignVectorCount !== 34
68
+ || manifest.qualification.runtimeActivation !== true || manifest.qualification.remainingConjunctsAreHostAuthenticated !== true) {
69
+ throw new Error("Structured producer manifest is not qualified for host content facts");
70
+ }
71
+ const producer = manifest.producers[0];
72
+ if (!exact(producer, ["id", "module", "digest", "authority", "activation", "kinds", "ruleDigests", "allowedImports",
73
+ "forbiddenCapabilities", "rules"]) || producer.id !== "structured-assurance-facts-v1"
74
+ || producer.module !== "adapters/common/structured-assurance-facts.mjs" || producer.digest !== sha(moduleBytes)
75
+ || producer.authority !== "none" || producer.activation !== "host-composite-content-v1"
76
+ || JSON.stringify(producer.kinds) !== JSON.stringify(["structured-log-claims", "config-document-literals", "policy-refusal-output"])
77
+ || !Array.isArray(producer.rules) || producer.rules.length !== 4) throw new Error("Structured producer identity is invalid");
78
+ const rules = new Map();
79
+ for (const rule of producer.rules) {
80
+ if (!exact(rule, ["id", "kind", "exportName", "ruleDigest", "definition"]) || !producer.kinds.includes(rule.kind)
81
+ || !producer.ruleDigests.includes(rule.ruleDigest) || rule.ruleDigest !== sha(JSON.stringify(rule.definition))
82
+ || rules.has(rule.ruleDigest)) throw new Error("Structured producer rule identity is invalid");
83
+ rules.set(rule.ruleDigest, rule);
84
+ }
85
+ return Object.freeze({ manifestDigest: sha(manifestBytes), producer, rules });
86
+ }
87
+ function pureObservation(item: CompositeItem, fact: any, snapshot: any, response: ResponseObservation) {
88
+ const rule = item.producer.rules.get(fact.ruleDigest), producer = item.producer.producer;
89
+ if (!rule || fact.producerId !== producer.id || fact.producerDigest !== producer.digest || fact.kind !== rule.kind) {
90
+ throw new Error("Composite fact is not bound to the installed producer rule");
91
+ }
92
+ const p = fact.parameters, d = rule.definition, materials = new Map(snapshot.materials.map((material: any) => [material.id, material]));
93
+ let result: any;
94
+ if (fact.kind === "structured-log-claims") {
95
+ if (p.responseFormat !== d.responseFormat || p.correlation !== d.correlation) throw new Error("Incident rule parameters changed");
96
+ result = compareIncidentClaims((materials.get(p.logMaterialId) as any)?.text, response.bytes, p.logMaterialId);
97
+ } else if (fact.kind === "config-document-literals") {
98
+ if (p.format !== d.format || JSON.stringify(p.requiredFields) !== JSON.stringify(d.requiredFields)) throw new Error("Document rule parameters changed");
99
+ result = compareConfigDocumentLiterals((materials.get(p.configMaterialId) as any)?.text,
100
+ (materials.get(p.documentMaterialId) as any)?.text);
101
+ } else if (fact.kind === "policy-refusal-output") {
102
+ if (p.rule !== d.rule || p.expectedDisposition !== d.expectedDisposition || p.responseFormat !== d.responseFormat
103
+ || p.nativeTemplateSetDigest !== d.nativeTemplateSetDigest) throw new Error("Refusal rule parameters changed");
104
+ result = comparePolicyRefusalOutput(p.rule, response.bytes, response.origin);
105
+ } else throw new Error("Composite fact has no pure structured producer");
106
+ if (!result || result.authority !== "none" || ![true, false, null].includes(result.matched)) throw new Error("Producer returned an invalid observation");
107
+ const observationDigest = sha(JSON.stringify({ factId: fact.id, result, materials: snapshot.identities,
108
+ responseDigest: sha(response.bytes) }));
109
+ if (result.matched === true) return { status: "pass", observationDigest, counterexampleRef: null, reasonCodes: [] };
110
+ if (result.matched === false) return { status: "fail", observationDigest,
111
+ counterexampleRef: sha(JSON.stringify(result)), reasonCodes: ["content-mismatch"] };
112
+ return { status: "unknown", observationDigest, counterexampleRef: null,
113
+ reasonCodes: [String(result.reason ?? "").startsWith("ambiguous-") ? "ambiguous-input" : "unsupported-input"] };
114
+ }
115
+ async function settleCompositeFact(item: CompositeItem, fact: any, collect: () => any): Promise<void> {
116
+ const reserved = item.authority.runtime.reserveFact(fact.id);
117
+ if (reserved.status === "current") { item.receipts.set(fact.id, reserved.receipt); return; }
118
+ if (reserved.status !== "reserved") throw new Error(`Composite fact ${fact.id} is ${reserved.status}`);
119
+ const receipt = await item.authority.runtime.settleFact({ reservation: reserved.reservation,
120
+ producer: item.authority.runtime.producerFor(fact.id), collect });
121
+ item.receipts.set(fact.id, receipt);
122
+ }
123
+ function exactResponse(observed: ReturnType<typeof persistedAssistantResponseObservation>, operationRef: string,
124
+ messageRequestId: string) {
125
+ return { origin: "assistant", entryId: observed.entryId, digest: observed.digest, byteLength: observed.byteLength,
126
+ operationRef, messageRequestId };
127
+ }
128
+ function settlementObservation(fact: any, response: any, matched: boolean, details: any,
129
+ reason: "persistence-mismatch" | "delivery-mismatch") {
130
+ return { status: matched ? "pass" : "fail", observationDigest: sha(JSON.stringify({ factId: fact.id, details })),
131
+ counterexampleRef: matched ? null : sha(JSON.stringify(details)), reasonCodes: matched ? [] : [reason], response };
132
+ }
133
+ function liveCompositeAssessment(item: CompositeItem, provisional: boolean, responseFile?: string) {
134
+ const decision = provisional ? { verdict: "pass", reasons: ["provisional-full-gate-preflight"] }
135
+ : item.authority.runtime.assess([...item.receipts.values()]);
136
+ const pass = decision.verdict === "pass";
137
+ return { version: "live-composite-assessment-v1", verdict: decision.verdict, completionAllowed: pass,
138
+ repairEligible: false, sourceMutationAllowed: false, assurance: pass ? "bounded-composite-contract-tested" : "none",
139
+ reasons: decision.reasons ?? [], failedChecks: decision.failedFactIds ?? [], missingChecks: decision.missingFactIds ?? [],
140
+ criterionId: item.contract.criterionId, criterionHash: item.contract.criterionHash,
141
+ sourcePath: item.authority.filePath, sourcePaths: [...new Set([item.authority.filePath, item.authority.anchorPath,
142
+ ...(responseFile ? [responseFile] : [])])],
143
+ factKinds: item.contract.facts.map((fact: any) => fact.kind) };
144
+ }
145
+ function persistedAssistantResponseObservationSafe(capability?: object) {
146
+ try { return capability ? persistedAssistantResponseObservation(capability) : undefined; }
147
+ catch { return undefined; }
148
+ }
149
+ function openCompositeAuthority(configuration: ReturnType<typeof openHostContractConfiguration>, contract: any,
150
+ binding: any, currentBinding: () => any) {
151
+ return configuration.withCompositeAuthority({ contract, binding, create: ({ key, directory, projectRoot, compiled,
152
+ contractText, declarations, maxAttempts }: any) => {
153
+ const sealedContext = sealCompositePlanContext({ key, contractText, declarations, binding, maxAttempts });
154
+ const identity = sha(JSON.stringify([binding.taskRunId, binding.sessionId, binding.criterionId,
155
+ binding.criterionHash, binding.runtimeGeneration, sealedContext.contextDigest]));
156
+ const filePath = path.join(directory, `composite-${identity}.jsonl`);
157
+ const journal = openCompositeAssuranceJournal({ filePath, projectRoot, key, contextDigest: sealedContext.contextDigest });
158
+ let closed = false;
159
+ try {
160
+ const runtime = openCompositeAssuranceRuntime({ key, sealedContext, journal, currentBinding: () => {
161
+ if (!configuration.isCurrent()) throw new Error("Composite host approval changed");
162
+ return currentBinding();
163
+ } });
164
+ const publicationStore = openCompositeTaskPublicationStore({ key, directory, projectRoot });
165
+ return Object.freeze({ compiled, sealedContext, runtime, filePath, anchorPath: journal.anchorPath, publicationStore,
166
+ close() { if (closed) return; closed = true; journal.close(); } });
167
+ } catch (error) { journal.close(); throw error; }
168
+ } });
169
+ }
170
+ /** Approved host configuration is opt-in; there is no model-authored oracle. */
171
+ export class IndependentAcceptanceRuntime {
172
+ readonly #options: Options;
173
+ readonly #entries = new Map<string, Entry>();
174
+ readonly #active = new Map<string, { token: symbol; generation: number }>();
175
+ readonly #retiring = new Map<string, Promise<void>>();
176
+ #nextGeneration = 0;
177
+ constructor(options: Options) { this.#options = options; }
178
+ #key(ctx: ExtensionContext, task: TaskContract) { return `${ctx.cwd}\0${task.sessionId}\0${task.taskRunId}`; }
179
+ #session(ctx: ExtensionContext) { return `${ctx.cwd}\0${ctx.sessionManager.getSessionId()}\0`; }
180
+ async #recoverPublication(ctx: ExtensionContext) {
181
+ const options = this.#options, task = options.activeTask(ctx); if (!options.configPath || !task) return;
182
+ let configuration: ReturnType<typeof openHostContractConfiguration>; try { configuration = openHostContractConfiguration({ configPath: options.configPath,
183
+ projectRoot: ctx.cwd, installedRoot: options.installedRoot }); } catch { return; }
184
+ try {
185
+ const approved = configuration.forRequest(task.operatorRequestDigest); if (!approved?.contracts.some((contract: any) => contract.route === "composite")) return;
186
+ const store = configuration.withCompositeRecovery((authority: any) => openCompositeTaskPublicationStore(authority));
187
+ const record: any = store.read(task.taskRunId, task.sessionId); if (!record) return;
188
+ if (!options.writeTask) throw new Error("Composite task publication writer is unavailable");
189
+ const completed = recoverCompositeTaskPublication({ store, record, task,
190
+ currentWorkingTreeDigest: workingTreeEvidenceDigest(workingTreeSnapshot(ctx.cwd)),
191
+ expectedCriteria: approved.contracts.filter((contract: any) => contract.route === "composite"),
192
+ cwd: ctx.cwd, writeTask: options.writeTask });
193
+ options.state.cacheTaskIdentity(ctx, completed);
194
+ } finally { configuration.close(); }
195
+ }
196
+ async activate(ctx: ExtensionContext): Promise<void> {
197
+ const prefix = this.#session(ctx);
198
+ while (this.#retiring.has(prefix)) await this.#retiring.get(prefix);
199
+ if (this.#active.has(prefix)) return;
200
+ this.#active.set(prefix, { token: Symbol("session-lifetime"), generation: ++this.#nextGeneration });
201
+ try { await this.#recoverPublication(ctx); } catch (error) { this.#active.delete(prefix); throw error; }
202
+ while (this.#active.size > 100) await this.#retireSession(this.#active.keys().next().value as string);
203
+ }
204
+ async #retire(entry: Entry): Promise<void> {
205
+ entry.block = "independent verification is stopping"; entry.controller.abort();
206
+ try { await entry.pending; } finally { entry.dispose(); entry.configuration?.close(); }
207
+ }
208
+ #retireSession(prefix: string): Promise<void> {
209
+ this.#active.delete(prefix);
210
+ const pending = this.#retiring.get(prefix); if (pending) return pending;
211
+ const retiring: Promise<void>[] = [];
212
+ for (const [key, entry] of this.#entries) if (key.startsWith(prefix)) {
213
+ this.#entries.delete(key); retiring.push(this.#retire(entry));
214
+ }
215
+ if (!retiring.length) return Promise.resolve();
216
+ const settled = Promise.all(retiring).then(() => {}).finally(() => {
217
+ if (this.#retiring.get(prefix) === settled) this.#retiring.delete(prefix);
218
+ });
219
+ this.#retiring.set(prefix, settled); return settled;
220
+ }
221
+ async prepare(ctx: ExtensionContext, task: TaskContract, response?: ResponseObservation): Promise<CompletionPreparation | false> {
222
+ const options = this.#options, prefix = this.#session(ctx), lifetime = this.#active.get(prefix);
223
+ const fence = () => {
224
+ try { return lifetime !== undefined && this.#active.get(prefix)?.token === lifetime.token && this.#session(ctx) === prefix
225
+ && task.sessionId === ctx.sessionManager.getSessionId() && options.activeTask(ctx)?.taskRunId === task.taskRunId; }
226
+ catch { return false; }
227
+ };
228
+ const ready = Object.freeze({ isCurrent: fence, completion: "ready" as const });
229
+ const deferred = (reason: string) => Object.freeze({ isCurrent: fence, completion: "deferred" as const, reason });
230
+ if (!fence()) return false;
231
+ if (!options.configPath || task.trace.outcome !== "pending") return ready;
232
+ const key = this.#key(ctx, task); let entry = this.#entries.get(key);
233
+ if (!entry) {
234
+ entry = { dispose: () => {}, runners: [], composites: [], hasComposite: false, controller: new AbortController() };
235
+ const owned = entry;
236
+ const unregister = registerIndependentAcceptanceProvider(ctx.cwd, task, {
237
+ read(_projected: TaskContract, digest: string, issue: (entry: any, assessment: any) => object) {
238
+ if (owned.controller.signal.aborted) return { block: "independent verification is stopping", stopReason: "stopping" };
239
+ if (owned.block) return { block: owned.block, stopReason: "approval" };
240
+ if (!owned.configuration?.isCurrent()) return { block: "independent host approval or installed verifier changed", stopReason: "approval" };
241
+ if (owned.stop) return { block: `independent verification is blocked: ${owned.stop.reason}`,
242
+ stopReason: owned.stop.reason, stopAttemptId: owned.stop.attemptId };
243
+ const actual = owned.composites.every(item => ["settled", "completed"].includes(item.authority.runtime.state().phase));
244
+ if (owned.hasComposite && !owned.provisional && !actual) return { block: owned.compositeReason
245
+ ?? "composite assurance settlement is pending", stopReason: "pending" };
246
+ const active = options.activeTask(ctx), snapshot = captureWorkspaceVerificationSnapshot(ctx.cwd);
247
+ const projectVerificationDigest = active && snapshot.digest === digest
248
+ ? options.state.projectVerification.currentDigest(ctx, active, snapshot, undefined, { completedProjection: true }) : null;
249
+ const entries = owned.runners.map(item => ({ ...item.contract, receipt: item.receipt }));
250
+ for (const item of owned.composites) { const assessment = liveCompositeAssessment(item,
251
+ owned.provisional === true, persistedAssistantResponseObservationSafe(owned.persistedResponse)?.file);
252
+ entries.push({ ...item.contract, assessment: issue(item.contract, assessment) }); }
253
+ return { entries, projectVerificationDigest };
254
+ },
255
+ webUiSettlementApplicability: () => owned.hasComposite ? "composite" : "not-applicable",
256
+ settleWebUi: (_projected: TaskContract, input: any) => this.#settleWebUi(ctx, task, owned, input)
257
+ });
258
+ owned.dispose = () => { unregister(); for (const item of owned.composites) {
259
+ try { if (!["completed", "cancelled", "superseded"].includes(item.authority.runtime.state().phase)) item.authority.runtime.invalidate("cancelled"); } catch {}
260
+ item.authority.close();
261
+ } };
262
+ this.#entries.set(key, owned);
263
+ while (this.#entries.size > 100) {
264
+ const oldest = this.#entries.keys().next().value as string, removed = this.#entries.get(oldest)!;
265
+ this.#entries.delete(oldest); await this.#retire(removed);
266
+ }
267
+ if (owned.controller.signal.aborted || !fence()) { if (this.#entries.get(key) === owned) this.#entries.delete(key); await this.#retire(owned); return false; }
268
+ try {
269
+ const configuration = openHostContractConfiguration({ configPath: options.configPath, projectRoot: ctx.cwd, installedRoot: options.installedRoot });
270
+ owned.configuration = configuration;
271
+ const approvedRequest = configuration.forRequest(task.operatorRequestDigest);
272
+ if (!approvedRequest || approvedRequest.nativeOnly === true) { owned.block = independentRequestAdmissionBlock(approvedRequest, task); return ready; }
273
+ for (const contract of approvedRequest.contracts) {
274
+ if (!task.acceptanceReceipt?.criteria.some(criterion => criterion.id === contract.criterionId && criterion.hash === contract.criterionHash)) {
275
+ throw new Error("Approved criterion mismatch");
276
+ }
277
+ if (contract.route === "composite") {
278
+ owned.hasComposite = true;
279
+ const plan = contract.planContext, compiledContract = JSON.parse(plan.contractText), declarations = JSON.parse(plan.declarationsText);
280
+ assertCompositeCriterion(task, compiledContract);
281
+ const producer = loadStructuredProducer(options.installedRoot);
282
+ const protectedPaths = new Set(plan.materialBindings.filter((binding: any) => binding.mode === "protected").map((binding: any) => binding.relativePath));
283
+ const currentBinding = () => {
284
+ const active = options.activeTask(ctx); if (!active || active.taskRunId !== task.taskRunId) throw new Error("Composite task is not current");
285
+ assertCompositeCriterion(active, compiledContract);
286
+ const snapshot = captureCompositeExecutionSnapshot({ projectRoot: ctx.cwd, materialBindings: plan.materialBindings, declarations });
287
+ if (loadStructuredProducer(options.installedRoot).manifestDigest !== producer.manifestDigest) throw new Error("Composite producer manifest changed");
288
+ const verification = captureWorkspaceVerificationSnapshot(ctx.cwd, { isProtectedProjectPath: (candidate: string) =>
289
+ protectedPaths.has((path.isAbsolute(candidate) ? path.relative(ctx.cwd, candidate) : candidate).split(path.sep).join("/")) });
290
+ const verifierDigest = compiledContract.facts.some((fact: any) => fact.kind === "project-verifier-current")
291
+ ? options.state.projectVerification?.currentDigest?.(ctx, active, verification) ?? null : null;
292
+ return { projectId: snapshot.projectId, projectHead: snapshot.projectHead, sourceDigest: snapshot.sourceDigest,
293
+ materialSnapshotDigest: snapshot.materialSnapshotDigest, suiteDigest: plan.identity.suiteDigest,
294
+ configDigest: plan.identity.configDigest, armDigest: plan.identity.armDigest, taskRunId: active.taskRunId,
295
+ sessionId: active.sessionId, operatorRequestDigest: active.operatorRequestDigest!, taskContractDigest: compositeTaskContractDigest(active),
296
+ criterionIndex: compiledContract.criterionIndex, criterionId: compiledContract.criterionId,
297
+ criterionHash: compiledContract.criterionHash, runtimeGeneration: lifetime!.generation,
298
+ phaseHeadDigest: compositePhaseHeadDigest(ctx.cwd, active), producerManifestDigest: producer.manifestDigest, verifierDigest };
299
+ };
300
+ const authority = openCompositeAuthority(configuration, contract, currentBinding(), currentBinding);
301
+ const recovered = authority.runtime.reissue();
302
+ if (recovered.completed) throw new Error("Completed composite journal has a nonterminal task contract");
303
+ const codePlans = new Map(authority.compiled.codePlans.map((item: any) => [item.digest, item])), codeChildren = new Map();
304
+ for (const fact of authority.compiled.contract.facts.filter((item: any) => item.kind === "bounded-code-checks")) {
305
+ const codePlan: any = codePlans.get(fact.parameters.codePlanDigest); if (!codePlan) throw new Error("Composite code child plan is missing");
306
+ codeChildren.set(fact.id, createCompositeCodeChildCollector({ state: options.state, context: ctx,
307
+ getTask: () => options.activeTask(ctx), approved: { store: configuration.store, ...approvedRequest.backend,
308
+ verifierDigest: approvedRequest.verifierDigest, authorizeSourceRead: ({ sourcePath }: any) =>
309
+ configuration.isCurrent() && options.authorizeSourceRead(ctx, sourcePath) },
310
+ plan: codePlan, fact, criterionId: contract.criterionId, criterionHash: contract.criterionHash,
311
+ maxAttempts: contract.maxAttempts }));
312
+ }
313
+ owned.composites.push({ contract: authority.compiled.contract, plan, declarations, producer, authority,
314
+ codeChildren,
315
+ receipts: new Map(recovered.facts.map((receipt: any) => [receipt.factId, receipt])),
316
+ ...(recovered.preparation ? { preparation: recovered.preparation } : {}),
317
+ ...(recovered.aggregate ? { aggregate: recovered.aggregate } : {}) });
318
+ continue;
319
+ }
320
+ const sourcePaths = new Set([contract.sourcePath, ...(contract.modulePaths ?? [])]);
321
+ const runner = createRuntimeContractRunner({ state: options.state, context: ctx, getTask: () => options.activeTask(ctx),
322
+ approved: { store: configuration.store, sourcePath: contract.sourcePath, modulePaths: contract.modulePaths, exportName: contract.exportName,
323
+ checks: contract.checks, ...approvedRequest.backend, verifierDigest: approvedRequest.verifierDigest,
324
+ authorizeSourceRead: ({ sourcePath }) => (!sourcePaths.has(sourcePath) || configuration.isCurrent()) && options.authorizeSourceRead(ctx, sourcePath) } });
325
+ owned.runners.push({ contract, runner });
326
+ }
327
+ } catch { owned.block = "independent host approval, composite identity, or installed producer is unavailable"; }
328
+ }
329
+ if (entry.controller.signal.aborted || !fence()) return false;
330
+ if (entry.block || !entry.configuration?.isCurrent()) return entry.hasComposite ? deferred(entry.block ?? "composite approval changed") : ready;
331
+ if (entry.pending) { await entry.pending; return !entry.controller.signal.aborted && fence()
332
+ ? entry.hasComposite ? deferred(entry.compositeReason ?? "composite settlement is pending") : ready : false; }
333
+ const owned = entry; owned.stop = undefined;
334
+ const pending = (async () => {
335
+ for (const item of owned.runners) {
336
+ if (owned.controller.signal.aborted) break;
337
+ try {
338
+ const result = await item.runner.run({ scope: { taskRunId: task.taskRunId, criterionId: item.contract.criterionId },
339
+ criterionHash: item.contract.criterionHash, maxAttempts: item.contract.maxAttempts, signal: owned.controller.signal });
340
+ const reason = ({ "evidence-pending": "pending", "evidence-interrupted": "interrupted", "evidence-exhausted": "exhausted" } as Record<string, string>)[result.reason];
341
+ const observed = result.evidence?.observed;
342
+ const executionThrew = observed !== null && typeof observed === "object" && "reason" in observed && observed.reason === "independent-execution-threw";
343
+ if (reason || executionThrew) { owned.stop = { reason: reason ?? "unavailable", attemptId: result.attemptId }; item.receipt = undefined; break; }
344
+ item.receipt = await item.runner.assess(result, { policy: "allow" });
345
+ } catch { item.receipt = undefined; owned.stop = { reason: "unavailable" }; break; }
346
+ }
347
+ if (!owned.hasComposite) return;
348
+ if (!response || response.origin !== "assistant" || typeof response.bytes !== "string" || !(response.bytes as any).isWellFormed()
349
+ || Buffer.byteLength(response.bytes) > 65_536) { owned.compositeReason = "composite actual assistant response is unavailable"; return; }
350
+ if (owned.response && (owned.response.origin !== response.origin || owned.response.bytes !== response.bytes)) {
351
+ owned.block = "a different composite response was already observed"; return;
352
+ }
353
+ owned.response = Object.freeze({ origin: response.origin, bytes: response.bytes });
354
+ for (const item of owned.composites) {
355
+ for (const fact of item.contract.facts.filter((candidate: any) => candidate.kind === "bounded-code-checks")) {
356
+ const child = item.codeChildren.get(fact.id); if (!child) throw new Error("Composite code child is unavailable");
357
+ await settleCompositeFact(item, fact, () => child.collect(task, owned.controller.signal));
358
+ }
359
+ const snapshot = captureCompositeExecutionSnapshot({ projectRoot: ctx.cwd,
360
+ materialBindings: item.plan.materialBindings, declarations: item.declarations });
361
+ for (const fact of item.contract.facts.filter((candidate: any) => item.producer.producer.kinds.includes(candidate.kind))) {
362
+ await settleCompositeFact(item, fact, async () => pureObservation(item, fact, snapshot, response));
363
+ }
364
+ const statuses = Object.values(item.authority.runtime.state().factStatuses);
365
+ owned.compositeReason = statuses.includes("error") ? "composite content producer failed"
366
+ : statuses.includes("fail") ? "composite content counterexample observed"
367
+ : "composite mediation, persistence, and terminal delivery facts are pending";
368
+ }
369
+ })();
370
+ owned.pending = pending;
371
+ try { await pending; } catch { owned.block = "independent composite producer execution is unavailable"; }
372
+ finally { if (owned.pending === pending) owned.pending = undefined; }
373
+ if (owned.controller.signal.aborted || !fence()) return false;
374
+ return owned.hasComposite ? deferred(owned.compositeReason ?? owned.block ?? "composite settlement is pending") : ready;
375
+ }
376
+ deferCompletion(ctx: ExtensionContext, task: TaskContract, response: ResponseObservation,
377
+ finalizer: DeferredCompletionFinalizer): boolean {
378
+ const entry = this.#entries.get(this.#key(ctx, task));
379
+ if (!entry?.hasComposite || entry.controller.signal.aborted || !entry.response
380
+ || response.origin !== entry.response.origin || response.bytes !== entry.response.bytes
381
+ || typeof finalizer?.preflight !== "function" || typeof finalizer?.publication !== "function"
382
+ || typeof finalizer?.finalize !== "function") return false;
383
+ if (entry.finalizer && entry.finalizer !== finalizer) return false;
384
+ entry.finalizer = finalizer; return true;
385
+ }
386
+ projectLifecycle(ctx: ExtensionContext, task: TaskContract, candidate: TaskContract): TaskContract | false {
387
+ const entry = this.#entries.get(this.#key(ctx, task));
388
+ if (!entry?.hasComposite || entry.controller.signal.aborted || !entry.configuration?.isCurrent()
389
+ || candidate.taskRunId !== task.taskRunId || candidate.sessionId !== task.sessionId) return false;
390
+ const ready = entry.composites.every(item => { const state = item.authority.runtime.state();
391
+ return ["settled", "completed"].includes(state.phase) || entry.provisional === true && item.preparation
392
+ && item.contract.facts.filter((fact: any) => fact.stage === "content").every((fact: any) => state.factStatuses[fact.id] === "pass"); });
393
+ return ready ? projectCompositeLifecycle(candidate, entry.composites.flatMap(item => item.contract.facts.map((fact: any) => fact.kind))) : false;
394
+ }
395
+ observeTurnEnd(ctx: ExtensionContext, message: any): void {
396
+ try {
397
+ const task = this.#options.activeTask(ctx); if (!task) return;
398
+ const entry = this.#entries.get(this.#key(ctx, task));
399
+ if (!entry?.hasComposite || !entry.response || entry.persistedResponse || entry.controller.signal.aborted) return;
400
+ entry.persistedResponse = capturePersistedAssistantResponse({ cwd: ctx.cwd, sessionId: task.sessionId,
401
+ manager: ctx.sessionManager, message, expectedText: entry.response.bytes });
402
+ entry.persistenceError = undefined;
403
+ } catch (error) {
404
+ const task = this.#options.activeTask(ctx), entry = task && this.#entries.get(this.#key(ctx, task));
405
+ if (entry?.hasComposite) entry.persistenceError = error instanceof Error ? error.message : "native response persistence unavailable";
406
+ }
407
+ }
408
+ async #settleWebUi(ctx: ExtensionContext, originalTask: TaskContract, owned: Entry, input: any) {
409
+ const blocked = (reason: string) => ({ status: "blocked", reason });
410
+ try {
411
+ if (owned.pending) await owned.pending;
412
+ const active = this.#options.activeTask(ctx);
413
+ if (!active || active.taskRunId !== originalTask.taskRunId || active.sessionId !== originalTask.sessionId
414
+ || active.trace.outcome !== "pending" || owned.controller.signal.aborted || !owned.configuration?.isCurrent()) {
415
+ return blocked("composite task authority is not current");
416
+ }
417
+ if (!input || typeof input !== "object" || !ID.test(input.operationRef) || !ID.test(input.messageRequestId)
418
+ || input.manager !== ctx.sessionManager || typeof input.evidence !== "function") {
419
+ return blocked("native WebUI settlement evidence is invalid");
420
+ }
421
+ if (!owned.persistedResponse || owned.persistenceError || !owned.response || !owned.finalizer) {
422
+ return blocked("native response persistence or completion finalizer is unavailable");
423
+ }
424
+ const persisted = persistedAssistantResponseObservation(owned.persistedResponse);
425
+ const response = exactResponse(persisted, input.operationRef, input.messageRequestId);
426
+ const brokerEvidence = await input.evidence();
427
+ const captures = owned.composites.map(item => ({ item, snapshot: captureCompositeExecutionSnapshot({
428
+ projectRoot: ctx.cwd, materialBindings: item.plan.materialBindings, declarations: item.declarations }) }));
429
+ const workspace = workingTreeSnapshot(ctx.cwd) as Record<string, string>;
430
+ for (const { item, snapshot } of captures) {
431
+ const mediation = openScopedMediationEvidence(brokerEvidence, { projectRoot: ctx.cwd, task: active,
432
+ operationRef: input.operationRef, messageRequestId: input.messageRequestId, plan: item.plan,
433
+ contract: item.contract, materials: snapshot.materials });
434
+ for (const fact of item.contract.facts.filter((candidate: any) => ["tool-policy-complete", "context-current",
435
+ "workspace-scope", "project-verifier-current"].includes(candidate.kind))) {
436
+ await settleCompositeFact(item, fact, async () => scopedMediationFactObservation(mediation,
437
+ { fact, plan: item.plan, contract: item.contract, task: active, materials: snapshot.materials, workspace }));
438
+ }
439
+ if (item.authority.runtime.assessContent([...item.receipts.values()]).verdict !== "pass") {
440
+ return blocked("composite content or mediation facts did not pass");
441
+ }
442
+ item.preparation = item.authority.runtime.prepareResponse({ facts: [...item.receipts.values()],
443
+ bytes: owned.response.bytes, origin: owned.response.origin, entryId: persisted.entryId,
444
+ operationRef: input.operationRef, messageRequestId: input.messageRequestId });
445
+ const persistenceFact = item.contract.facts.find((fact: any) => fact.kind === "response-persisted");
446
+ if (!persistenceFact) return blocked("composite response persistence fact is missing");
447
+ await settleCompositeFact(item, persistenceFact, async () => settlementObservation(persistenceFact, response, true,
448
+ { entryId: persisted.entryId, digest: persisted.digest, byteLength: persisted.byteLength,
449
+ fileDigest: persisted.fileDigest }, "persistence-mismatch"));
450
+ }
451
+ let finalization: object | false, target: any;
452
+ owned.provisional = true;
453
+ try { finalization = owned.finalizer.preflight(); target = finalization && owned.finalizer.publication(finalization); }
454
+ finally { owned.provisional = false; }
455
+ if (!finalization || !target || target.version !== "composite-terminal-task-target-v1"
456
+ || target.terminalTaskDigest !== compositeTaskPublicationDigest(target.terminalTask)) {
457
+ return blocked("full completion preflight did not pass");
458
+ }
459
+ const deliveryKey = sha(JSON.stringify(["piagent-composite-terminal-delivery-v1", active.taskRunId,
460
+ active.sessionId, input.operationRef, input.messageRequestId, persisted.entryId, persisted.digest]));
461
+ const delivered = commitCompositeWebUiDelivery({ response: owned.persistedResponse, manager: input.manager,
462
+ taskRunId: active.taskRunId, operationRef: input.operationRef, messageRequestId: input.messageRequestId,
463
+ idempotencyKey: deliveryKey });
464
+ const delivery = webUiDeliveryObservation(delivered);
465
+ for (const { item } of captures) {
466
+ const deliveryFact = item.contract.facts.find((fact: any) => fact.kind === "terminal-delivery");
467
+ if (!deliveryFact) return blocked("composite terminal delivery fact is missing");
468
+ await settleCompositeFact(item, deliveryFact, async () => settlementObservation(deliveryFact, response, true,
469
+ { operationRef: delivery.operationRef, messageRequestId: delivery.messageRequestId,
470
+ confirmationEntryId: delivery.confirmationEntryId, confirmationDigest: delivery.confirmationDigest },
471
+ "delivery-mismatch"));
472
+ }
473
+ for (const { item } of captures) item.aggregatePreparation = item.authority.runtime.prepareAggregatePublication({
474
+ preparation: item.preparation, facts: [...item.receipts.values()], taskPublicationDigest: target.terminalTaskDigest });
475
+ const publicationStore = captures[0]?.item.authority.publicationStore;
476
+ if (!publicationStore) return blocked("composite publication store is unavailable");
477
+ const publicationRecord = publicationStore.prepare({ pendingTask: target.pendingTask,
478
+ terminalTask: target.terminalTask, workingTreeDigest: target.workingTreeDigest,
479
+ entries: captures.map(({ item }) => item.aggregatePreparation.entry) });
480
+ for (const { item } of captures) {
481
+ item.aggregate = item.authority.runtime.settleAggregate({ publication: item.aggregatePreparation.publication });
482
+ item.publication = item.authority.runtime.prepareTaskCompletion({ aggregate: item.aggregate });
483
+ }
484
+ if (!owned.finalizer.finalize(finalization)) return blocked("terminal task publication failed");
485
+ const completed = this.#options.activeTask(ctx);
486
+ if (!completed || completed.taskRunId !== active.taskRunId || completed.sessionId !== active.sessionId
487
+ || completed.trace.outcome !== "completed" || !durableTaskContractMatches(ctx.cwd, completed)) {
488
+ return blocked("terminal task contract is not durably persisted");
489
+ }
490
+ for (const { item } of captures) item.authority.runtime.completeTask({ aggregate: item.aggregate,
491
+ publication: item.publication });
492
+ publicationStore.complete(publicationRecord);
493
+ const taskStatus = compositeSettlementTaskStatus(owned.composites.flatMap(item => item.contract.facts.map((fact: any) => fact.kind)));
494
+ owned.compositeReason = undefined; return { status: "settled", taskStatus };
495
+ } catch {
496
+ return blocked("composite settlement unavailable");
497
+ }
498
+ }
499
+ clear(ctx: ExtensionContext): Promise<void> { return this.#retireSession(this.#session(ctx)); }
500
+ }